grok-build-upstream-mirror/crates/codegen/xai-grok-pager/src/views/turn_status.rs

1688 lines
64 KiB
Rust
Raw Normal View History

//! Turn status line — single-row widget showing current turn activity.
//!
//! Layout: `⠧ Run command 0.2s 1m20s ⇣12k [stop]`
//!
//! - Spinner (left, slowed to ~7.5fps)
//! - Activity label (colored per activity type, truncates if needed)
//! - Phase timer `Xs` (gray, never truncates)
//! - Queued-send hint `· N queued — Enter to send now` (gray, sendable waits only)
//! - Fill space
//! - Turn timer `Xm Ys` and optional token count `⇣Nk` (right-aligned, gray)
//! - Cancel button `[stop]` (right-aligned, red on hover)
//!
//! Hidden when idle (0 height). Appears between scrollback and prompt.
use std::time::{Duration, Instant};
use ratatui::buffer::Buffer;
use ratatui::layout::Rect;
use ratatui::style::{Color, Modifier, Style};
use ratatui::text::{Line, Span};
use unicode_width::UnicodeWidthStr;
use xai_grok_workspace::permission::mcp_pretty_name_if_qualified;
use crate::acp::tracker::{TurnActivity, WaitingReason};
use crate::app::agent::{AgentCommand, AgentState};
use crate::app::agent_view::McpInitProgress;
use crate::render::line_utils::truncate_str;
use crate::theme::Theme;
/// Show each spinner frame for this many animation ticks.
/// At ~30fps, 4 ticks = ~133ms per frame = ~7.5 spinner fps.
pub(crate) const SPINNER_DIVISOR: u64 = 4;
/// Show each monitor-pulse frame for this many animation ticks — twice the
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
/// [`SPINNER_DIVISOR`] dwell (~3.75 fps). The idle still-running cue should
/// breathe calmly rather than read like the active turn spinner, so its
/// `○ ◎ ◉ ◎` cycle runs at roughly half the speed (~1.07s per loop).
pub(crate) const MONITOR_PULSE_DIVISOR: u64 = 8;
/// Pulse speed for every "waiting on you" diamond — the drain-blocked
/// status, the pending-user-input status, and the plan-approval status
/// all share this cadence. `pulse_brightness` returns `sin²(tick*speed)`,
/// which has period π, so at ~30fps this is ~1.3s per cycle
/// (`π / (0.08 * 30) ≈ 1.31`).
///
/// Always route diamond rendering through [`pending_diamond_color`] so
/// the three call sites can never silently drift apart.
pub(crate) const USER_WAITING_PULSE_SPEED: f32 = 0.08;
/// Compute the pulsing diamond color for any "waiting on you" cue.
///
/// Blends `accent` toward `theme.bg_base` using a `sin²` pulse driven by
/// [`USER_WAITING_PULSE_SPEED`]. Brightness ranges from 0.3 (dim) to 1.0
/// (full accent) so the diamond stays visible at the trough.
///
/// Pass `theme.accent_user` for user-input waits (permission prompts,
/// `ask_user_question`, the drain-blocked idle status) and
/// `theme.accent_plan` for plan-approval waits.
pub(crate) fn pending_diamond_color(theme: &Theme, accent: Color, tick: u64) -> Color {
let brightness = crate::theme::pulse_brightness(tick, USER_WAITING_PULSE_SPEED);
crate::render::color::blend_color(theme.bg_base, accent, 0.3 + brightness * 0.7)
.unwrap_or(accent)
}
// ---------------------------------------------------------------------------
// Output
// ---------------------------------------------------------------------------
/// Output from rendering the turn status line.
#[derive(Debug, Default)]
pub struct TurnStatusOutput {
/// Hit area for the cancel button, if rendered.
/// `None` when the button is not shown (idle, cancelling, drain-blocked).
pub cancel_button: Option<Rect>,
/// Hit area for the background-demote button, if rendered.
pub bg_button: Option<Rect>,
2026-07-24 16:59:42 +00:00
/// Hit area for the still-running watcher cue (click opens the tasks
/// pane). `None` on keyboard-only hosts.
pub watching_cue: Option<Rect>,
}
2026-07-24 16:59:42 +00:00
/// Hover state for the turn-status row's mouse affordances (`[stop]`, `[↓]`,
/// the still-running watcher cue). `Some(_)` renders them; `None` marks a
/// keyboard-only host (minimal mode — no mouse capture) and suppresses all.
#[derive(Debug, Clone, Copy, Default)]
pub struct MouseButtons {
/// Whether the mouse is over the `[stop]` cancel button.
pub cancel_hovered: bool,
/// Whether the mouse is over the `[↓]` send-to-background button.
pub bg_hovered: bool,
2026-07-24 16:59:42 +00:00
/// Whether the mouse is over the still-running watcher cue.
pub watching_hovered: bool,
}
/// Counts of idle-surviving "watcher" work — background jobs that can wake
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
/// the agent for a new turn while it sits idle (commands and monitors on
/// completion/events, `/loop` tasks on a timer, background subagents on
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
/// finish). They share one persistent still-running cue above the prompt.
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
/// Broader than the tasks-pane `Watchers` group (monitors + loops only).
#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
pub struct Watchers {
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
/// Running background commands (non-monitor `background: true` tasks).
pub commands: usize,
/// Running `monitor` background tasks.
pub monitors: usize,
/// Active scheduled `/loop` tasks.
pub loops: usize,
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
/// Running background subagents. While the agent is idle, any running
/// subagent is a background one — a foreground subagent would keep the
/// parent in `TurnRunning`.
pub subagents: usize,
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
pub workflows: usize,
}
impl Watchers {
/// Total watcher count across all kinds.
pub fn total(self) -> usize {
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
self.commands + self.monitors + self.loops + self.subagents + self.workflows
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
}
/// Awaitable in-flight work — the kinds a blocking `wait_tasks` /
/// `get_task_output` wait can resolve on (commands, monitors, subagents;
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
/// scheduled `/loop` tasks and workflows are not task waits).
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
pub fn awaitable_work(self) -> usize {
self.commands + self.monitors + self.subagents
}
}
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
/// Format a counts-first `"… still running"` cue from `(count, noun)` pairs,
/// listing only the non-zero kinds (plain-`s` plurals) — e.g.
/// `"1 command · 2 monitors still running"`. `None` when every count is
/// zero. Single owner of the format mechanics so the agent view's idle cue
/// and the dashboard's background-work label cannot drift.
pub(crate) fn format_still_running<'a>(
kinds: impl IntoIterator<Item = (usize, &'a str)>,
) -> Option<String> {
use std::fmt::Write as _;
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
let mut label = String::with_capacity(48);
for (count, noun) in kinds {
if count == 0 {
continue;
}
if !label.is_empty() {
label.push_str(" \u{00b7} ");
}
let plural = if count == 1 { "" } else { "s" };
let _ = write!(label, "{count} {noun}{plural}");
}
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
if label.is_empty() {
return None;
}
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
label.push_str(" still running");
Some(label)
}
/// The idle watcher cue's label — e.g.
/// `"1 command · 2 monitors · 1 loop · 1 subagent still running"`. Leads
/// with the counts (not an ambient "watching") so a glance under a
/// "Worked for X" marker still reads as unfinished work. `None` when no
/// watchers are live.
fn still_running_label(watchers: Watchers) -> Option<String> {
format_still_running([
(watchers.commands, "command"),
(watchers.monitors, "monitor"),
(watchers.loops, "loop"),
(watchers.subagents, "subagent"),
(watchers.workflows, "workflow"),
])
}
/// Whether the turn is blocked in a wait the shell aborts as soon as the
/// user sends a message (`get_task_output` with `timeout_ms`, `wait_tasks`,
/// `Await*`, and a foreground subagent await — mirrors the shell's blocking
/// waits, whose send-now routing cancels the blocked turn and runs the new
/// message next). Typing is actionable during these, which is what the
/// parked-wait rendering (`AgentView::is_parked_on_sendable_wait` /
/// `renders_parked`) builds on.
///
/// `Subagent` is included: the shell treats a blocked foreground subagent
/// await like the other blocking waits, so Enter sends promptly and pre-wait
/// rows read as held. `Model` waits stay excluded — the model is actively
/// producing the turn, so a message typed there queues behind real work. Pure
/// predicate over the resolved activity; no turn-lifecycle side effects.
pub fn is_sendable_wait(activity: &Option<TurnActivity>) -> bool {
matches!(
activity,
Some(TurnActivity::Waiting(
WaitingReason::TaskOutput { waits: true, .. }
| WaitingReason::TasksComplete
| WaitingReason::Sleep
| WaitingReason::Subagent
))
)
}
2026-07-24 16:59:42 +00:00
/// Inputs to [`render_turn_status`] — one frame's worth of turn state.
#[derive(Debug)]
pub struct TurnStatusArgs<'a> {
pub state: &'a AgentState,
pub activity: &'a Option<TurnActivity>,
pub turn_elapsed: Option<Duration>,
pub activity_started_at: Option<Instant>,
pub tick: u64,
pub drain_blocked: bool,
/// Mouse affordances + hover state; `None` for keyboard-only hosts.
pub buttons: Option<MouseButtons>,
pub has_running_execute: bool,
/// Context-window tokens used, shown as `⇣Nk`.
pub total_tokens: Option<u64>,
pub mcp_init_progress: Option<&'a McpInitProgress>,
pub is_bash_turn: bool,
pub is_pending_user_input: bool,
pub goal_verifying: bool,
pub watchers: Watchers,
/// Parked on a sendable wait (`AgentView::renders_parked`).
2026-07-24 16:59:42 +00:00
pub parked: bool,
/// Transparent right-side background so the row blends with the
/// terminal's own background (minimal mode).
pub flat_background: bool,
pub held_queue: usize,
pub held_queue_top_sendable: bool,
}
/// Render the turn status line into the given area.
///
/// The caller is responsible for only allocating a 1-row area when
/// `should_show()` returns true (and 0 rows when false).
pub fn render_turn_status(
buf: &mut Buffer,
area: Rect,
2026-07-24 16:59:42 +00:00
args: TurnStatusArgs<'_>,
) -> TurnStatusOutput {
2026-07-24 16:59:42 +00:00
let TurnStatusArgs {
state,
activity,
turn_elapsed,
activity_started_at,
tick,
drain_blocked,
buttons,
has_running_execute,
total_tokens,
mcp_init_progress,
is_bash_turn,
is_pending_user_input,
goal_verifying,
watchers,
parked,
flat_background,
held_queue,
held_queue_top_sendable,
} = args;
// Resolve the mouse affordances: a keyboard-only host (`None`) suppresses
// both buttons and reports no hover.
let show_buttons = buttons.is_some();
let cancel_hovered = buttons.is_some_and(|b| b.cancel_hovered);
let bg_hovered = buttons.is_some_and(|b| b.bg_hovered);
if area.height == 0 || area.width < 10 {
return TurnStatusOutput::default();
}
let theme = Theme::current();
// MCP startup seed (total == 0) while idle — show "Starting session…"
// above the prompt until the shell reports real server counts. Real MCP
// progress (total > 0) renders as the compact top-bar chip instead, not
// here. Auto-expires via `is_visible()` if the shell never reports.
if state.is_idle()
&& !drain_blocked
&& let Some(progress) = mcp_init_progress
&& progress.total == 0
&& progress.is_visible()
{
render_starting_session(buf, area, progress, tick, &theme);
return TurnStatusOutput::default();
}
// Special case: drain is blocked (user editing front prompt, agent idle).
// No cancel button in this state.
if drain_blocked && state.is_idle() {
// Pulsing diamond in accent_user, blending toward bg.
let diamond_color = pending_diamond_color(&theme, theme.accent_user, tick);
let spans = vec![
Span::styled(
format!("{} ", crate::glyphs::diamond_filled()),
Style::default().fg(diamond_color),
),
Span::styled(
"agent idle ~ waiting on your edit",
Style::default().fg(theme.gray),
),
];
buf.set_line(area.x, area.y, &Line::from(spans), area.width);
return TurnStatusOutput::default();
}
// Idle or parked: persistent cue (not scrollback — it must never scroll
// away). Lower priority than the starting-session and drain-blocked cues
// above. Parked never falls through to the running-turn chrome
// (spinner/timers/[stop]) — the wait aborts the moment the user types,
// so that chrome would lie.
if state.is_idle() || parked {
// Parked with held queued rows: the queued hint IS the input-semantics
// story (Enter acts on the queue immediately), so it replaces the
// generic interrupt copy.
let parked_suffix = if held_queue > 0 && held_queue_top_sendable {
format!(" \u{00b7} {held_queue} queued — Enter to send now")
} else if held_queue > 0 {
format!(" \u{00b7} {held_queue} queued")
2026-07-24 16:59:42 +00:00
} else {
" \u{00b7} send a message to interrupt".to_string()
2026-07-24 16:59:42 +00:00
};
let cue = match (still_running_label(watchers), parked) {
(Some(label), true) => Some(format!("{label}{parked_suffix}")),
(Some(label), false) => Some(label),
(None, true) => Some(format!("waiting{parked_suffix}")),
(None, false) => None,
2026-07-24 16:59:42 +00:00
};
if let Some(cue) = cue {
// Pulsing concentric circle (○ ◎ ◉ ◎) on a calm ambient cadence:
// the agent is idle, so this breath runs slower than the active
// turn spinner (see MONITOR_PULSE_DIVISOR).
let frames = crate::glyphs::monitor_icon_frames();
let frame_idx = (tick / MONITOR_PULSE_DIVISOR) as usize % frames.len();
let icon = format!("{} ", frames[frame_idx]);
let label_fg = if buttons.is_some_and(|b| b.watching_hovered) {
theme.text_primary
} else {
theme.gray
};
let cue_width = (icon.width() + cue.width()).min(area.width as usize) as u16;
let spans = vec![
Span::styled(icon, Style::default().fg(theme.accent_system)),
Span::styled(cue, Style::default().fg(label_fg)),
];
buf.set_line(area.x, area.y, &Line::from(spans), area.width);
// The cue opens the tasks pane on click — only advertise the hit
// area when there are tasks to show (a watcherless parked cue has
// nothing behind it).
return TurnStatusOutput {
watching_cue: (show_buttons && watchers.total() > 0)
.then(|| Rect::new(area.x, area.y, cue_width, 1)),
..TurnStatusOutput::default()
};
}
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
return TurnStatusOutput::default();
}
// Determine if cancel button should be shown.
// Show when: TurnRunning or CommandRunning.
// Hide when: Idle, Cancelling (already cancelling), or a keyboard-only host
// (no clickable buttons — see `buttons`).
let show_cancel = show_buttons
&& matches!(
state,
AgentState::TurnRunning | AgentState::CommandRunning { .. }
);
// ── Compute activity style and label ──
let (activity_style, label, is_tool) =
compute_activity(&theme, state, activity, is_bash_turn, goal_verifying);
// Early return for idle (shouldn't happen if should_show is respected, but be safe).
if matches!(state, AgentState::Idle) {
return TurnStatusOutput::default();
}
// ── Build right-aligned content first (to know how much space is left) ──
// Format: `1m20s` or `1m20s ⇣12k` (with tokens).
let turn_timer_str = match (turn_elapsed, total_tokens) {
(Some(d), Some(tokens)) if tokens > 0 => {
format!(
"{} {}{}",
format_turn_timer(d),
crate::glyphs::token_arrow(),
format_tokens_short(tokens)
)
}
(Some(d), _) => format_turn_timer(d),
_ => String::new(),
};
let turn_timer_width = turn_timer_str.width();
// Bg button: [↓] normally, [send to bg] when hovered (only for running execute
// tools). `show_cancel` already implies a mouse host, so no extra check.
let show_bg = show_cancel && has_running_execute;
let bg_str = if show_bg {
if bg_hovered {
" [send to bg]"
} else {
" [\u{2193}]"
}
} else {
""
};
let bg_width = bg_str.width();
// Cancel button: always `[stop]`. Leading space only when the bg button
// is not shown (otherwise they're adjacent). Every arm is a `&'static str`
// so the per-frame status line never allocates. Hover state is conveyed by
// color (red on hover, see `cancel_style`), not by swapping the label.
let cancel_str: &str = match (show_cancel, show_bg) {
(false, _) => "",
(true, true) => "[stop]",
(true, false) => " [stop]",
};
let cancel_width = cancel_str.width();
let right_width = turn_timer_width + bg_width + cancel_width;
// ── Build components ──
// While a tool is blocked on a permission prompt or `ask_user_question`,
// swap the running braille spinner for a pulsing `◆`. Same animation
// shape the drain-blocked and plan-approval indicators already use,
// so every "your turn" status reads with one consistent visual cue.
let spinner_str = if is_pending_user_input {
format!("{} ", crate::glyphs::diamond_filled())
} else {
let frames = crate::glyphs::braille_spinner_frames();
let frame_idx = (tick / SPINNER_DIVISOR) as usize % frames.len();
format!("{} ", frames[frame_idx])
};
let spinner_width = spinner_str.width();
// "Ask" tools (AskUserQuestion): suppress the phase timer so the user
// doesn't feel time-pressured while answering questions.
let is_asking = is_tool
&& matches!(
activity,
Some(TurnActivity::ToolRunning { title, .. })
if title.starts_with("Ask: ") || title.starts_with("Ask ")
);
// Phase timer (gray, same as turn timer) — hidden for ask tools
let phase_timer_str = if is_asking {
String::new()
} else {
activity_started_at
.map(|t| format!(" {}", format_turn_timer(t.elapsed())))
.unwrap_or_default()
};
let phase_timer_width = phase_timer_str.width();
// Timer style (gray for both phase and turn timers).
//
// Right-side elements (turn timer, bg button, cancel button) must set
// fg, bg, AND remove_modifier explicitly. fill_background() paints
// bg_base on every cell before widgets render, but set_line() for the
// left content may overwrite fg/modifiers on cells in the right zone.
// A Style with bg:None (the default) cannot restore bg after a reset,
// and a Style without remove_modifier cannot clear leaked modifiers.
// Right-side cells normally paint `bg_base`; in a flat-background host
// (minimal mode) use the terminal's own background so the row stays
// transparent like the rest of the live region.
let timer_bg = if flat_background {
Color::Reset
} else {
theme.bg_base
};
let timer_style = Style::default()
.fg(theme.gray)
.bg(timer_bg)
.remove_modifier(Modifier::all());
// Available width for activity label (only the label truncates)
// Layout: spinner + label + phase_timer + queued_hint + gap(1) + turn_timer + cancel
let min_gap = 1;
let available_for_label = (area.width as usize)
.saturating_sub(spinner_width)
.saturating_sub(phase_timer_width)
.saturating_sub(min_gap)
.saturating_sub(right_width);
// ── Render left side: spinner + label (truncated) + phase_timer + queued_hint ──
let mut left_spans: Vec<Span<'static>> = Vec::with_capacity(5);
// Spinner color: usually inherits the activity color (green for tools,
// secondary for thinking/responding, yellow for retries). While the
// tool is parked on the user we render `◆` with a smooth pulse from
// dim→bright in `accent_user`, matching the drain-blocked and
// plan-approval indicators so every "your turn" status has the same
// visual cadence.
let spinner_style = if is_pending_user_input {
let diamond_color = pending_diamond_color(&theme, theme.accent_user, tick);
Style::default().fg(diamond_color)
} else {
activity_style
};
left_spans.push(Span::styled(spinner_str, spinner_style));
// Activity label (potentially truncated)
let mut queued_hint: Option<Span<'static>> = None;
if is_tool {
if let Some(TurnActivity::ToolRunning { title, description }) = activity {
if is_asking {
// Ask tools: render as a unified gray label (like Thinking/Responding),
// not as a command invocation — yellow is reserved for shell commands.
let detail = title
.strip_prefix("Ask: ")
.or_else(|| title.strip_prefix("Ask "))
.unwrap_or(title.as_str());
let msg = format!("Waiting on answers for {detail}");
let display = truncate_str(&msg, available_for_label);
left_spans.push(Span::styled(display, activity_style));
} else if let Some(desc) = description
.as_deref()
.map(str::trim)
.filter(|s| !s.is_empty())
{
// Bash (and similar) tools carry a human description — prefer
// that over the raw command for the status line so a sleep /
// long-running exec reads as `{description}…` rather than
// `Run sleep 5 && …`.
let msg = crate::acp::tracker::format_waiting_for_subject(desc);
let display = truncate_str(&msg, available_for_label);
left_spans.push(Span::styled(display, activity_style));
} else if let Some(query) = title.strip_prefix("Web search: ") {
// Web search: "Search " (muted) + query (yellow)
let prefix = "Search ";
let prefix_width = prefix.width();
let query = query.trim_matches('"');
let max_query = available_for_label.saturating_sub(prefix_width).max(5);
let display = truncate_str(query, max_query);
left_spans.push(Span::styled(prefix, Style::default().fg(theme.gray)));
left_spans.push(Span::styled(display, Style::default().fg(theme.command)));
} else if let Some(url) = title.strip_prefix("Fetch: ") {
// Fetch tools: "Fetch " (muted) + URL (yellow)
let prefix = "Fetch ";
let prefix_width = prefix.width();
let max_url = available_for_label.saturating_sub(prefix_width).max(5);
let display = truncate_str(url, max_url);
left_spans.push(Span::styled(prefix, Style::default().fg(theme.gray)));
left_spans.push(Span::styled(display, Style::default().fg(theme.command)));
} else {
// Normal tool: "Run " (muted) + command (syntax-highlighted).
// For qualified MCP tool names the activity title is the
// raw `server__action` string from ACP; prettify it to
// `(Server) Action` so the spinner doesn't show the ugly
// delimiter form. Non-MCP titles (bash commands etc.) are
// returned untouched by `mcp_pretty_name_if_qualified`.
let prefix = "Run ";
let pretty = mcp_pretty_name_if_qualified(title.as_str());
let detail = pretty.as_str();
let prefix_width = prefix.width();
let max_cmd = available_for_label.saturating_sub(prefix_width).max(5);
let first_line = detail.lines().next().unwrap_or(detail);
let display = truncate_str(first_line, max_cmd);
left_spans.push(Span::styled(prefix, Style::default().fg(theme.gray)));
left_spans.extend(crate::views::tasks_pane::highlight_bash_command(&display));
}
}
} else {
// Sendable wait holding queued messages: the persistent inline hint
// saying why the queue is paused and how to send anyway. On the status
// row (not an ephemeral tip) so it stays visible for the whole wait,
// and dropped before the label truncates on a narrow terminal.
// "Enter to send now" is advertised only when Enter would actually
// send the top row (bash / client-expanded local rows refuse with a
// toast — see `AgentView::held_queue_top_sendable`).
let suffix = if held_queue > 0 && is_sendable_wait(activity) {
if held_queue_top_sendable {
format!(" · {held_queue} queued — Enter to send now")
} else {
format!(" · {held_queue} queued")
}
} else {
String::new()
};
if !suffix.is_empty() && label.width() + suffix.width() <= available_for_label {
left_spans.push(Span::styled(label.clone(), activity_style));
queued_hint = Some(Span::styled(suffix, Style::default().fg(theme.gray)));
} else {
let display = truncate_str(&label, available_for_label);
left_spans.push(Span::styled(display, activity_style));
}
}
// Phase timer (gray, never truncates)
if !phase_timer_str.is_empty() {
left_spans.push(Span::styled(phase_timer_str, timer_style));
}
// After the phase timer, so the elapsed time reads as the wait's, not the hint's.
if let Some(hint) = queued_hint {
left_spans.push(hint);
}
// Render left side
let left_line = Line::from(left_spans);
buf.set_line(area.x, area.y, &left_line, area.width);
// ── Render right side: turn_timer + bg + cancel ──
let right_start_x = area.x + area.width.saturating_sub(right_width as u16);
// Helper: build a fully-specified right-side style (fg + bg + clear mods).
let right_style = |fg| {
Style::default()
.fg(fg)
.bg(timer_bg)
.remove_modifier(Modifier::all())
};
// Turn timer (gray)
let mut x = right_start_x;
if !turn_timer_str.is_empty() {
let span = Span::styled(turn_timer_str.clone(), timer_style);
buf.set_span(x, area.y, &span, turn_timer_width as u16);
x += turn_timer_width as u16;
}
// Bg button — accent_running on hover
let bg_button_rect = if show_bg && !bg_str.is_empty() {
let bg_x = x;
let bg_style = if bg_hovered {
right_style(theme.accent_running)
} else {
right_style(theme.gray)
};
let span = Span::styled(bg_str, bg_style);
buf.set_span(x, area.y, &span, bg_width as u16);
x += bg_width as u16;
Some(Rect::new(bg_x, area.y, bg_str.width() as u16, 1))
} else {
None
};
// Cancel button — accent_error (red) on hover, gray at rest
let cancel_button_rect = if show_cancel && !cancel_str.is_empty() {
let cancel_x = x;
let cancel_style = if cancel_hovered {
right_style(theme.accent_error)
} else {
right_style(theme.gray)
};
let span = Span::styled(cancel_str, cancel_style);
buf.set_span(x, area.y, &span, cancel_width as u16);
Some(Rect::new(cancel_x, area.y, cancel_width as u16, 1))
} else {
None
};
TurnStatusOutput {
cancel_button: cancel_button_rect,
bg_button: bg_button_rect,
2026-07-24 16:59:42 +00:00
watching_cue: None,
}
}
/// Compute activity style, label, and whether it's a tool.
fn compute_activity(
theme: &Theme,
state: &AgentState,
activity: &Option<TurnActivity>,
is_bash_turn: bool,
goal_verifying: bool,
) -> (Style, String, bool) {
match (state, activity) {
(AgentState::TurnCancelling | AgentState::CommandCancelling { .. }, _) => (
Style::default().fg(theme.accent_error),
"Cancelling…".to_string(),
false,
),
// Goal-mode completion verification runs in-turn after the model
// stops streaming. The harness drives the skeptic panel (the model
// itself is idle), but the turn's last streaming activity can still
// read as `Responding`/`Thinking`; label the whole window
// "Verifying…" so the multi-minute panel isn't mislabelled as the
// model responding (or a hung "Waiting…").
(AgentState::TurnRunning, _) if goal_verifying => (
Style::default().fg(theme.text_secondary),
"Verifying…".to_string(),
false,
),
(AgentState::TurnRunning, Some(TurnActivity::Thinking)) => (
Style::default().fg(theme.text_secondary),
"Thinking…".to_string(),
false,
),
(AgentState::TurnRunning, Some(TurnActivity::Responding)) => (
Style::default().fg(theme.text_secondary),
"Responding…".to_string(),
false,
),
(AgentState::TurnRunning, Some(TurnActivity::ToolRunning { title, description })) => {
// "Ask" tools (AskUserQuestion) use gray spinner like Thinking —
// green feels out of place when the user is answering questions.
// Human descriptions (e.g. bash `description`) also use muted
// secondary — they read as a wait subject (`Wait 5s…`), not a
// green `Run <command>` invocation.
let is_ask = title.starts_with("Ask: ") || title.starts_with("Ask ");
let has_desc = description
.as_deref()
.map(str::trim)
.is_some_and(|s| !s.is_empty());
let style = if is_ask || has_desc {
Style::default().fg(theme.text_secondary)
} else {
Style::default().fg(theme.accent_success)
};
(style, String::new(), true)
}
(AgentState::TurnRunning, Some(TurnActivity::AutoCompacting)) => (
Style::default().fg(theme.text_secondary),
"Compacting…".to_string(),
false,
),
(AgentState::TurnRunning, Some(TurnActivity::Retrying { attempt, .. })) => (
Style::default().fg(theme.warning),
format!("Retrying (attempt {attempt})…"),
false,
),
(AgentState::TurnRunning, Some(TurnActivity::Waiting(reason))) => (
// Explicit wait reason (model / subagent / task output / tasks /
// sleep): name what the agent is blocked on instead of a generic
// "Waiting…". See `WaitingReason` and `AgentView::resolve_turn_activity`.
Style::default().fg(theme.text_secondary),
reason.label(),
false,
),
(AgentState::TurnRunning, None) if is_bash_turn => (
// Bash turn: not inference, show generic "Running…".
Style::default().fg(theme.text_secondary),
"Running…".to_string(),
false,
),
(AgentState::TurnRunning, None) => (
// Fallback: a running inference turn with no resolved activity. The
// view resolves this gap into Waiting(Model/Subagent) before render,
// so this is now a rarely-hit safety net.
Style::default().fg(theme.text_secondary),
"Waiting…".to_string(),
false,
),
(
AgentState::CommandRunning {
command:
command @ (AgentCommand::CreateWorktree
| AgentCommand::RestoreWorktree
| AgentCommand::RestoreCode
| AgentCommand::ForkSession),
..
},
_,
) => (
Style::default().fg(theme.gray),
format!("{}", command.display_name()),
false,
),
(AgentState::CommandRunning { command, .. }, _) => (
Style::default().fg(theme.text_secondary),
format!("{}", command.display_name()),
false,
),
(AgentState::Idle, _) => (Style::default(), String::new(), false),
}
}
/// Whether the idle "Starting session…" indicator wants the turn-status row.
///
/// True only for a fresh `total == 0` startup seed (gated by
/// [`McpInitProgress::is_visible`] so an orphaned seed expires). Real MCP
/// progress (`total > 0`) renders as the top-bar chip instead, so it does not
/// drive this row.
fn starting_session_visible(progress: Option<&McpInitProgress>) -> bool {
progress.is_some_and(|p| p.total == 0 && p.is_visible())
}
/// Render the idle "Starting session…" indicator above the prompt.
///
/// Format: `⠋ Starting session… 0:01` — braille spinner + label + elapsed
/// timer. Rendered in `theme.gray_dim` (the dimmest gray) so it reads as
/// quiet/ambient, matching the top-bar MCP chip and the directory path — this
/// is non-blocking startup, not foreground activity. Shown only while the MCP
/// init progress is a startup seed (`total == 0`), before the shell reports
/// real server counts; real progress (`total > 0`) renders as the top-bar chip.
fn render_starting_session(
buf: &mut Buffer,
area: Rect,
progress: &McpInitProgress,
tick: u64,
theme: &Theme,
) {
let frames = crate::glyphs::braille_spinner_frames();
let frame_idx = (tick / SPINNER_DIVISOR) as usize % frames.len();
let timer_str = format!(" {}", format_turn_timer(progress.started_at.elapsed()));
let style = Style::default().fg(theme.gray_dim);
let spans = vec![
Span::styled(format!("{} ", frames[frame_idx]), style),
Span::styled("Starting session…", style),
Span::styled(timer_str, style),
];
buf.set_line(area.x, area.y, &Line::from(spans), area.width);
}
/// Whether the turn status line should be visible.
///
/// Returns true when a turn is active (Running or Cancelling), when the drain
/// is blocked (agent idle, waiting on user edit), while the MCP startup seed
/// is showing "Starting session…" (a fresh `total == 0` seed), or when the
/// agent is idle but background watchers are still running
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
/// (`watchers.total() > 0`) — running commands and monitors wake the agent on
/// completion/events, scheduled `/loop` tasks fire prompts, and background
/// subagents inject a completion turn, any of which can start a new turn.
///
/// A parked turn always shows the row, watchers or not.
///
/// Real MCP progress (`total > 0`) renders as a compact chip in the top status
/// bar instead, so it does not affect this row.
pub fn should_show(
state: &AgentState,
drain_blocked: bool,
mcp_init_progress: Option<&McpInitProgress>,
watchers: Watchers,
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
parked: bool,
) -> bool {
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
if parked {
return true;
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
}
!state.is_idle()
|| drain_blocked
|| starting_session_visible(mcp_init_progress)
|| watchers.total() > 0
}
/// Format a duration for the turn/phase timer.
///
/// Re-exports [`crate::util::format_duration`] under the old name for
/// backwards compatibility within this module.
pub use crate::util::format_duration as format_turn_timer;
/// Format a token count for compact display.
///
/// - Under 1000: `1`, `10`, `100` (raw number)
/// - 1k-100k: `1.23k`, `10.1k` (with decimal)
/// - 100k-1m: `100k`, `500k` (whole thousands)
/// - 1m+: `1.23m`, `10.1m` (with decimal)
fn format_tokens_short(tokens: u64) -> String {
if tokens < 1000 {
format!("{tokens}")
} else if tokens < 100_000 {
// 1k-99.9k: show one or two decimals for precision
let k = tokens as f64 / 1000.0;
if tokens < 10_000 {
format!("{k:.2}k") // 1.23k
} else {
format!("{k:.1}k") // 10.1k
}
} else if tokens < 1_000_000 {
// 100k-999k: whole thousands
let k = tokens / 1000;
format!("{k}k")
} else {
// 1m+: show with decimal
let m = tokens as f64 / 1_000_000.0;
if tokens < 10_000_000 {
format!("{m:.2}m") // 1.23m
} else {
format!("{m:.1}m") // 10.1m
}
}
}
#[cfg(test)]
mod tests {
use std::time::Duration;
use super::*;
/// Sendable waits = exactly the wait reasons the shell aborts on a queued
/// user prompt (blocking task-output / wait_tasks / Await, and a blocked
/// foreground subagent await — all take the send-now path). Model waits —
/// where typing only queues behind the actively-streaming turn — and
/// non-wait activities keep the busy spinner.
#[test]
fn sendable_wait_matches_shell_interruptible_waits() {
let task_wait = |waits| {
Some(TurnActivity::Waiting(WaitingReason::TaskOutput {
task_ids: vec!["t-1".into()],
subject: Some("sleep 300".into()),
waits,
}))
};
assert!(is_sendable_wait(&task_wait(true)));
assert!(
!is_sendable_wait(&task_wait(false)),
"instant polls are not blocking waits"
);
assert!(is_sendable_wait(&Some(TurnActivity::Waiting(
WaitingReason::TasksComplete
))));
assert!(is_sendable_wait(&Some(TurnActivity::Waiting(
WaitingReason::Sleep
))));
assert!(!is_sendable_wait(&Some(TurnActivity::Waiting(
WaitingReason::Model
))));
assert!(
is_sendable_wait(&Some(TurnActivity::Waiting(WaitingReason::Subagent))),
"the shell aborts a blocked foreground subagent await on send-now, \
so Enter during it must read as sendable"
);
assert!(!is_sendable_wait(&Some(TurnActivity::Thinking)));
assert!(!is_sendable_wait(&None));
}
#[test]
fn format_subsecond() {
assert_eq!(format_turn_timer(Duration::from_millis(500)), "0.5s");
assert_eq!(format_turn_timer(Duration::from_millis(120)), "0.1s");
}
#[test]
fn format_under_10s_has_decimal() {
assert_eq!(format_turn_timer(Duration::from_secs_f64(5.2)), "5.2s");
assert_eq!(format_turn_timer(Duration::from_secs_f64(9.9)), "9.9s");
}
#[test]
fn format_10s_plus_no_decimal() {
assert_eq!(format_turn_timer(Duration::from_secs(10)), "10s");
assert_eq!(format_turn_timer(Duration::from_secs(32)), "32s");
assert_eq!(format_turn_timer(Duration::from_secs(59)), "59s");
}
#[test]
fn format_minutes() {
assert_eq!(format_turn_timer(Duration::from_secs(60)), "1m0s");
assert_eq!(format_turn_timer(Duration::from_secs(80)), "1m20s");
assert_eq!(format_turn_timer(Duration::from_secs(600)), "10m0s");
}
#[test]
fn activity_label_reads_verifying_while_goal_verifying_overriding_stale_activity() {
let theme = Theme::current();
// Running turn, no streaming activity, goal verifying → "Verifying…".
let (_, label, _) = compute_activity(&theme, &AgentState::TurnRunning, &None, false, true);
assert_eq!(label, "Verifying…");
// Same state without the verifying flag → generic "Waiting…".
let (_, label, _) = compute_activity(&theme, &AgentState::TurnRunning, &None, false, false);
assert_eq!(label, "Waiting…");
// During verification the model is idle but its last streaming
// activity (Responding/Thinking) can linger — the flag overrides it
// so the panel reads "Verifying…", not "Responding…" (the bug).
for activity in [TurnActivity::Responding, TurnActivity::Thinking] {
let (_, label, _) = compute_activity(
&theme,
&AgentState::TurnRunning,
&Some(activity),
false,
true,
);
assert_eq!(label, "Verifying…");
}
// Without the flag the streaming label stands.
let (_, label, _) = compute_activity(
&theme,
&AgentState::TurnRunning,
&Some(TurnActivity::Responding),
false,
false,
);
assert_eq!(label, "Responding…");
}
#[test]
fn waiting_reason_renders_specific_label() {
use crate::acp::tracker::WaitingReason;
let theme = Theme::current();
let cases = [
(WaitingReason::Model, "Waiting for response…"),
(WaitingReason::Subagent, "Waiting on subagent…"),
(WaitingReason::task_output(), "Waiting on task output…"),
(
WaitingReason::TaskOutput {
task_ids: vec!["t1".into()],
subject: Some("compile release".into()),
waits: false,
},
"compile release…",
),
(WaitingReason::TasksComplete, "Waiting on tasks…"),
(WaitingReason::Sleep, "Sleeping…"),
];
for (reason, expected) in cases {
let (_, label, is_tool) = compute_activity(
&theme,
&AgentState::TurnRunning,
&Some(TurnActivity::Waiting(reason.clone())),
false,
false,
);
assert_eq!(label, expected, "reason {reason:?}");
assert!(!is_tool, "waiting is not a tool activity");
}
}
#[test]
fn bash_turn_still_renders_running_not_waiting() {
let theme = Theme::current();
// A bash (non-inference) turn with no activity keeps its own "Running…"
// label — the view leaves it as `None` rather than Waiting(Model).
let (_, label, _) = compute_activity(&theme, &AgentState::TurnRunning, &None, true, false);
assert_eq!(label, "Running…");
}
#[test]
fn format_hours() {
assert_eq!(format_turn_timer(Duration::from_secs(3600)), "1h0m");
assert_eq!(format_turn_timer(Duration::from_secs(3725)), "1h2m");
}
#[test]
fn should_show_when_running() {
assert!(should_show(
&AgentState::TurnRunning,
false,
None,
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
Watchers::default(),
false
));
assert!(should_show(
&AgentState::TurnCancelling,
false,
None,
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
Watchers::default(),
false
));
assert!(!should_show(
&AgentState::Idle,
false,
None,
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
Watchers::default(),
false
));
}
#[test]
fn should_show_when_drain_blocked() {
assert!(should_show(
&AgentState::Idle,
true,
None,
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
Watchers::default(),
false
));
}
#[test]
fn should_show_when_watchers_running() {
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
// Idle but a watcher (command, monitor, loop, or subagent) is still
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
// running → row stays visible so the persistent "… still running" cue
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
// can show.
for watchers in [
Watchers {
commands: 1,
..Watchers::default()
},
Watchers {
monitors: 1,
..Watchers::default()
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
},
Watchers {
loops: 1,
..Watchers::default()
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
},
Watchers {
subagents: 1,
..Watchers::default()
},
] {
assert!(should_show(&AgentState::Idle, false, None, watchers, false));
}
// Idle with no watchers and nothing else pending → hidden.
assert!(!should_show(
&AgentState::Idle,
false,
None,
Watchers::default(),
false
));
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
}
#[test]
fn should_show_parked_always() {
assert!(should_show(
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
&AgentState::TurnRunning,
false,
None,
Watchers {
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
commands: 1,
..Watchers::default()
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
},
true
));
assert!(should_show(
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
&AgentState::TurnRunning,
false,
None,
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
Watchers::default(),
true
));
}
#[test]
fn should_show_when_starting_session() {
// A fresh total == 0 seed shows "Starting session…" above the prompt.
let seed = McpInitProgress {
total: 0,
connected: 0,
started_at: Instant::now(),
};
assert!(should_show(
&AgentState::Idle,
false,
Some(&seed),
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
Watchers::default(),
false
));
// Real progress (total > 0) is the top-bar chip — it must NOT drive
// this row.
let connecting = McpInitProgress {
total: 3,
connected: 1,
started_at: Instant::now(),
};
assert!(!should_show(
&AgentState::Idle,
false,
Some(&connecting),
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
Watchers::default(),
false
));
// An expired seed must not drive the row either.
let expired = McpInitProgress {
total: 0,
connected: 0,
started_at: Instant::now() - McpInitProgress::SEED_EXPIRE - Duration::from_secs(1),
};
assert!(!should_show(
&AgentState::Idle,
false,
Some(&expired),
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
Watchers::default(),
false
));
}
/// Collect every rendered glyph in `area` into a single string.
fn buffer_text(buf: &Buffer, area: Rect) -> String {
(area.y..area.y + area.height)
.map(|y| {
(area.x..area.x + area.width)
.filter_map(|x| buf.cell((x, y)).map(|c| c.symbol().to_string()))
.collect::<String>()
})
.collect::<Vec<_>>()
.join("\n")
}
2026-07-24 16:59:42 +00:00
/// Baseline render args: idle agent on a mouse host with the given watchers.
fn idle_args<'a>(watchers: Watchers) -> TurnStatusArgs<'a> {
TurnStatusArgs {
state: &AgentState::Idle,
activity: &None,
turn_elapsed: None,
activity_started_at: None,
tick: 0,
drain_blocked: false,
buttons: Some(MouseButtons::default()),
has_running_execute: false,
total_tokens: None,
mcp_init_progress: None,
is_bash_turn: false,
is_pending_user_input: false,
goal_verifying: false,
watchers,
parked: false,
flat_background: false,
held_queue: 0,
held_queue_top_sendable: false,
}
}
/// Render `args` into a `width`×1 row.
fn render_row(args: TurnStatusArgs<'_>, width: u16) -> (TurnStatusOutput, Buffer) {
let area = Rect::new(0, 0, width, 1);
let mut buf = Buffer::empty(area);
let output = render_turn_status(&mut buf, area, args);
(output, buf)
}
/// Render `args` into a `width`×1 row, returning the visible text.
fn render_row_text(args: TurnStatusArgs<'_>, width: u16) -> String {
let (_, buf) = render_row(args, width);
buffer_text(&buf, buf.area)
}
/// Invoke `render_turn_status` for an idle agent with the given MCP seed.
fn render_idle_with_mcp(progress: &McpInitProgress) -> String {
2026-07-24 16:59:42 +00:00
let mut args = idle_args(Watchers::default());
args.mcp_init_progress = Some(progress);
render_row_text(args, 60)
}
/// Invoke `render_turn_status` for an idle agent with the given watcher
/// counts at animation tick `tick`.
fn render_idle_with_watchers_at_tick(watchers: Watchers, tick: u64) -> String {
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
render_idle_with_watchers_in_width(watchers, tick, 72)
}
/// [`render_idle_with_watchers_at_tick`] with an explicit row width.
fn render_idle_with_watchers_in_width(watchers: Watchers, tick: u64, width: u16) -> String {
2026-07-24 16:59:42 +00:00
let mut args = idle_args(watchers);
args.tick = tick;
render_row_text(args, width)
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
}
/// Invoke `render_turn_status` for a PARKED running turn (the stopped
/// look) with the given watcher counts.
fn render_parked_with_watchers(watchers: Watchers) -> String {
2026-07-24 16:59:42 +00:00
let activity = Some(TurnActivity::Waiting(WaitingReason::TasksComplete));
let mut args = idle_args(watchers);
args.state = &AgentState::TurnRunning;
args.activity = &activity;
args.turn_elapsed = Some(Duration::from_secs(5));
args.parked = true;
render_row_text(args, 72)
}
/// Invoke `render_turn_status` for an idle agent with the given watcher
/// counts at the first animation tick.
fn render_idle_with_watchers(watchers: Watchers) -> String {
render_idle_with_watchers_at_tick(watchers, 0)
}
/// Invoke `render_turn_status` for an idle agent with `n` running
/// monitors at animation tick `tick`.
fn render_idle_with_monitors_at_tick(n: usize, tick: u64) -> String {
render_idle_with_watchers_at_tick(
Watchers {
monitors: n,
..Watchers::default()
},
tick,
)
}
/// Invoke `render_turn_status` for an idle agent with `n` running
/// monitors at the first animation tick.
fn render_idle_with_monitors(n: usize) -> String {
render_idle_with_monitors_at_tick(n, 0)
}
#[test]
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
fn idle_with_monitors_renders_still_running_cue() {
let text = render_idle_with_monitors(2);
assert!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
text.contains("2 monitors still running"),
"idle with monitors must render the still-running cue, got: {text:?}"
);
}
#[test]
fn idle_with_one_monitor_uses_singular() {
let text = render_idle_with_monitors(1);
assert!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
text.contains("1 monitor still running") && !text.contains("monitors"),
"single monitor must use the singular noun, got: {text:?}"
);
}
#[test]
fn idle_with_no_monitors_renders_nothing() {
let text = render_idle_with_monitors(0);
assert!(
text.trim().is_empty(),
"idle with no monitors must render nothing, got: {text:?}"
);
}
2026-07-24 16:59:42 +00:00
/// Mouse hosts get a hit rect hugging exactly the rendered cue text, and
/// hover brightens the label; keyboard-only hosts get neither.
#[test]
fn watching_cue_is_clickable_on_mouse_hosts_only() {
let theme = Theme::current();
let watchers = Watchers {
monitors: 1,
..Watchers::default()
};
// First label cell (after the 2-col icon).
let label_fg = |buf: &Buffer| buf.cell((2, 0)).map(|c| c.fg);
let (output, buf) = render_row(idle_args(watchers), 60);
let rect = output.watching_cue.expect("mouse host must get a hit rect");
let rendered_width = buffer_text(&buf, buf.area).trim_end().width() as u16;
assert_eq!(rect, Rect::new(0, 0, rendered_width, 1));
assert_eq!(label_fg(&buf), Some(theme.gray));
let mut args = idle_args(watchers);
args.buttons = Some(MouseButtons {
watching_hovered: true,
..MouseButtons::default()
});
let (_, buf) = render_row(args, 60);
assert_eq!(label_fg(&buf), Some(theme.text_primary));
let mut args = idle_args(watchers);
args.buttons = None;
let (output, _) = render_row(args, 60);
assert!(output.watching_cue.is_none());
}
#[test]
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
fn idle_with_loops_renders_still_running_cue() {
let text = render_idle_with_watchers(Watchers {
loops: 2,
..Watchers::default()
});
assert!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
text.contains("2 loops still running"),
"idle with loops must render the still-running cue, got: {text:?}"
);
}
#[test]
fn idle_with_one_loop_uses_singular() {
let text = render_idle_with_watchers(Watchers {
loops: 1,
..Watchers::default()
});
assert!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
text.contains("1 loop still running") && !text.contains("loops"),
"single loop must use the singular noun, got: {text:?}"
);
}
#[test]
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
fn idle_with_subagents_renders_still_running_cue() {
let text = render_idle_with_watchers(Watchers {
subagents: 2,
..Watchers::default()
});
assert!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
text.contains("2 subagents still running"),
"idle with subagents must render the still-running cue, got: {text:?}"
);
}
#[test]
fn idle_with_one_subagent_uses_singular() {
let text = render_idle_with_watchers(Watchers {
subagents: 1,
..Watchers::default()
});
assert!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
text.contains("1 subagent still running") && !text.contains("subagents"),
"single subagent must use the singular noun, got: {text:?}"
);
}
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
#[test]
fn idle_with_one_workflow_counts_run_once() {
let text = render_idle_with_watchers(Watchers {
workflows: 1,
..Watchers::default()
});
assert!(text.contains("1 workflow still running"), "got: {text:?}");
}
#[test]
fn idle_with_monitors_and_loops_lists_both() {
// Both watcher kinds present → one cue lists monitors then loops,
// each with its own count, joined by the middle-dot separator.
let text = render_idle_with_watchers(Watchers {
monitors: 1,
loops: 2,
..Watchers::default()
});
assert!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
text.contains("1 monitor \u{00b7} 2 loops still running"),
"both kinds must be listed in one cue, got: {text:?}"
);
}
#[test]
fn idle_with_all_watcher_kinds_lists_all() {
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
// Commands, monitors, loops, and subagents present → one cue lists
// all four in order, middle-dot separated.
let text = render_idle_with_watchers(Watchers {
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
commands: 1,
monitors: 2,
loops: 1,
subagents: 3,
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
workflows: 0,
});
assert!(
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
text.contains(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
"1 command \u{00b7} 2 monitors \u{00b7} 1 loop \u{00b7} 3 subagents still running"
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
),
"all kinds must be listed in one cue, got: {text:?}"
);
}
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
#[test]
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
fn narrow_area_clips_cue_tail_keeping_counts() {
// 40 cols with three kinds: the row tail-clips with no ellipsis, so
// the leading counts survive and the trailing suffix is what gets
// cut. Pins the narrow-pane tradeoff of leading with the counts; a
// smarter compact fallback would be a behavior change.
let watchers = Watchers {
commands: 1,
monitors: 2,
loops: 1,
..Watchers::default()
};
let text = render_idle_with_watchers_in_width(watchers, 0, 40);
assert!(
text.contains("1 command \u{00b7} 2 monitors \u{00b7} 1 loop"),
"the counts must survive the clip, got: {text:?}"
);
}
#[test]
fn idle_with_commands_renders_still_running_cue() {
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
// Plain background commands (non-monitor bg tasks) count as watchers:
// they wake the agent with a task-completed turn, so the cue must show.
let text = render_idle_with_watchers(Watchers {
commands: 2,
..Watchers::default()
});
assert!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
text.contains("2 commands still running"),
"idle with bg commands must render the still-running cue, got: {text:?}"
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
);
let text = render_idle_with_watchers(Watchers {
commands: 1,
..Watchers::default()
});
assert!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
text.contains("1 command still running") && !text.contains("commands"),
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
"single command must use the singular noun, got: {text:?}"
);
}
#[test]
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
fn parked_with_watchers_renders_cue_not_running_chrome() {
// The wait aborts as soon as the user types, so busy chrome would lie.
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
let text = render_parked_with_watchers(Watchers {
commands: 2,
..Watchers::default()
});
assert!(
text.contains("2 commands still running \u{00b7} send a message to interrupt"),
"parked with bg work must render the interruptible still-running cue, got: {text:?}"
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
);
assert!(
!text.contains("Waiting") && !text.contains("[stop]"),
"parked must not render the running-turn chrome, got: {text:?}"
);
}
#[test]
fn parked_without_watchers_renders_waiting_cue() {
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
let text = render_parked_with_watchers(Watchers::default());
assert!(
text.contains("waiting \u{00b7} send a message to interrupt"),
"watcherless parked must render the waiting interrupt cue, got: {text:?}"
);
assert!(
!text.contains("[stop]"),
"watcherless parked must not render the running-turn chrome, got: {text:?}"
);
}
#[test]
fn parked_with_held_queue_renders_queued_hint() {
// The queued hint replaces the interrupt copy (Enter = send-now).
let activity = Some(TurnActivity::Waiting(WaitingReason::TasksComplete));
let mut args = idle_args(Watchers {
commands: 1,
..Watchers::default()
});
args.state = &AgentState::TurnRunning;
args.activity = &activity;
args.parked = true;
args.held_queue = 1;
args.held_queue_top_sendable = true;
let text = render_row_text(args, 80);
assert!(
text.contains("1 queued — Enter to send now"),
"parked with a held row must advertise the queued hint, got: {text:?}"
);
assert!(
!text.contains("send a message to interrupt"),
"queued hint replaces the interrupt copy, got: {text:?}"
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
);
}
#[test]
fn idle_with_no_watchers_renders_nothing() {
let text = render_idle_with_watchers(Watchers::default());
assert!(
text.trim().is_empty(),
"idle with no watchers must render nothing, got: {text:?}"
);
}
#[test]
fn queued_hint_renders_after_phase_timer() {
2026-07-24 16:59:42 +00:00
let activity = Some(TurnActivity::Waiting(WaitingReason::Subagent));
let mut args = idle_args(Watchers::default());
args.state = &AgentState::TurnRunning;
args.activity = &activity;
args.activity_started_at = Some(Instant::now() - Duration::from_secs(359));
args.held_queue = 1;
args.held_queue_top_sendable = true;
let text = render_row_text(args, 80);
assert!(
text.contains("Waiting on subagent… 5m59s · 1 queued — Enter to send now"),
"phase timer must sit between the wait label and the queued hint, got: {text:?}"
);
}
#[test]
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
fn still_running_label_lists_only_nonzero_kinds() {
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
assert_eq!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
still_running_label(Watchers {
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
commands: 2,
..Watchers::default()
}),
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
Some("2 commands still running".into())
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
);
assert_eq!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
still_running_label(Watchers {
monitors: 2,
..Watchers::default()
}),
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
Some("2 monitors still running".into())
);
assert_eq!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
still_running_label(Watchers {
loops: 1,
..Watchers::default()
}),
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
Some("1 loop still running".into())
);
assert_eq!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
still_running_label(Watchers {
subagents: 1,
..Watchers::default()
}),
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
Some("1 subagent still running".into())
);
assert_eq!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
still_running_label(Watchers {
monitors: 1,
loops: 2,
..Watchers::default()
}),
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
Some("1 monitor \u{00b7} 2 loops still running".into())
);
assert_eq!(
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
still_running_label(Watchers {
Synced from monorepo Changes: - Gate session-lifecycle heap steady state with a dhat soak - Unbreak merge lifecycle e2e after default model → grok-4.5 - Scan home-scope rules dirs at <root>/rules - Complete text-input paste and terminal parity - Gate project roles and personas - Use canonical editing in dialogs - Use canonical editing in search bars - Reject ambiguous MCP tool IDs - Harden Git operands for plugins - Simplify queue drain API - Pass RFC 9207 iss through MCP OAuth token exchange - Show leader roster when local agents map is empty - Use canonical editing in Persona views - Remove marketplace default-skills auto-install and purge old installs - Use canonical editing in extension forms - Add canonical dashboard text editing - Use canonical editing in settings - Add /summarize as a /recap alias - Restore previous agent when exiting dashboard - Use tool_choice auto for compaction - Settings toggle for snap-prompt-to-top on send - Update default models to grok-4.5 - Source login shell once for local bash (env + alias/function snapshot) - Template hardcoded param names in server-native tool descriptions - Fix System-Reminder XML tag injection in CLAUDE.md via agents_md - Fix remote workspace-server hardcoding LSP trust (repo code execution risk) - Clear orphaned tool-call updates at turn end - Suppress task wake after cancel - Send x-grok-client-identifier on direct API tool calls - Harden dashboard peek lease transitions - Host /btw side panel in live region (minimal mode) - Bound scroll presentation latency - Highlight multi-line constructs correctly in diffs and the file viewer - Block web_fetch non-public IPs; local opt-in is explicit-host only - Seed coding_data_retention_opt_out=false for OAuth e2es in pty-harness - Follow up clipboard delivery feedback - Use canonical editing in pickers - Route TextArea through canonical editor - Persistent "watching" status row; quieter turn markers - Gate sensitive edit targets - Expose agent registry counts and gate session churn on them - Default coding data sharing to opt-out until server preference applies - Wire chat attachment ids through gateway prompts - On auth refresh failure, issue retry - Forward preview provenance and computer lifecycle state - Document independent privacy controls and scope /privacy output - Strip SamplingError Display prefix on rate-limit UI copy - Stop dumping Cloudflare HTML into Retry failed - Disable in-place prompt edit (scroll jank on enter) - Strip forced ANSI color from gh pr view JSON - Plumb bash tool description onto ToolUsageCard wire
2026-07-18 19:48:28 +01:00
commands: 1,
monitors: 1,
loops: 1,
subagents: 2,
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
workflows: 0,
}),
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
Some(
"1 command \u{00b7} 1 monitor \u{00b7} 1 loop \u{00b7} 2 subagents still running"
.into()
)
);
Synced from monorepo Synced from monorepo Changes: - Shell: accept target response id on rewind execute - Shell: stamp response id on chat user message chunks - Worktree: optional rebuild and stale git registration cleanup in auto-GC - Worktree: kind-aware auto-GC TTLs and config knobs - Worktree: macOS process CWD scan and Unix PID liveness for GC guards - Worktree: automatic throttled GC on startup (Linux age-based; non-Linux dead-only) - Pager: add `[ui].combine_queued_prompts` to batch queued follow-ups - Shell: stop overwriting user skills - Tools: read markdown in `skills/` directories untruncated - `/usage` shows per-session token and dollar usage in the TUI - Security: prompt on environment-dumping `ps` variants - Security: always-safe `kubectl` no longer runs arbitrary kubeconfig credential plugins without permission - Tools: make scheduler deletion durable - Shell: add relocation storage primitives - Shell: give side model calls their own conversation ids - Fix five workflow-runtime bugs (budget, pause, cancel, reconnect) - Security: peel `env -S` / `--split-string` operands in the Bash permission gate (managed deny/ask) - Pager: expose doctor in the TUI - Security: block unauthorized RCE via abused safe commands - Pager idle watcher cue: "1 subagent still running" instead of "watching · 1 subagent" - Security: block `rg --pre` arbitrary code execution in auto-mode - Voice: diagnose silent-mic failures (macOS permission) and add doctor/terminal-setup Voice section - App builder deployer: `allow_forking` and `show_built_with_grok` - Pager: stop stacking duplicate "Worked for" markers on parked turns - Shell: support `max` as a distinct reasoning effort tier - Tools: serialize background `/loop` fires on the whole work unit - Shell: add working-directory relocation state primitives - Proto: `ClientToolResult` and `ChatConfig` client-side tools - Shell: model providers - Chat: select App Builder product on the Build path - Shell: attach author identity to feedback when the deployment opts in - Doctor: fix for SSH wrap setup - Workflow authoring skills: create-workflow and import-claude-workflow docs - Add read-only grok doctor - Sandbox: apply Landlock without a controlling TTY - Pager: recover image paste over grok wrap on headless remotes - Pager: make actions screen-mode aware - Shell: resume sessions when the working directory moves - Pager: centralize terminal diagnostics - Workspace: gate inline shell file access - Pager: centralize terminal probes - Pager: edit minimal prompts in an external editor - Pager: standardize backgrounding on Ctrl+B - Shell: recap rides the parent turn's prompt cache - Tools: add scheduler lifecycle version clock Source-Revision: 0f4d7c91b8b2b408333f6de1e8a76cb8eaa71899
2026-07-21 18:10:23 +00:00
assert_eq!(still_running_label(Watchers::default()), None);
}
#[test]
fn idle_monitor_icon_animates_across_ticks() {
// The leading glyph cycles through monitor_icon_frames() as `tick`
// advances, so two ticks a full frame apart (0 vs MONITOR_PULSE_DIVISOR)
// must render different icons — proving the cue is animated, not static.
let frame0 = render_idle_with_monitors_at_tick(1, 0);
let frame1 = render_idle_with_monitors_at_tick(1, MONITOR_PULSE_DIVISOR);
let icon0 = frame0.chars().next();
let icon1 = frame1.chars().next();
assert_ne!(
icon0, icon1,
"monitor icon must animate between frames, got {frame0:?} vs {frame1:?}"
);
}
#[test]
fn idle_zero_server_seed_renders_starting_session() {
// total == 0 seed → "Starting session…" above the prompt.
let text = render_idle_with_mcp(&McpInitProgress {
total: 0,
connected: 0,
started_at: Instant::now(),
});
assert!(
text.contains("Starting session"),
"idle 0-server seed must render 'Starting session…', got: {text:?}"
);
}
#[test]
fn idle_active_mcp_progress_renders_nothing_in_turn_status() {
// total > 0 is the top-bar chip — the turn-status row stays empty.
let text = render_idle_with_mcp(&McpInitProgress {
total: 3,
connected: 1,
started_at: Instant::now(),
});
assert!(
text.trim().is_empty(),
"active MCP progress must NOT render in the turn-status row, got: {text:?}"
);
}
#[test]
fn expired_seed_renders_nothing() {
// An expired total == 0 seed renders nothing — defense-in-depth.
let text = render_idle_with_mcp(&McpInitProgress {
total: 0,
connected: 0,
started_at: Instant::now() - McpInitProgress::SEED_EXPIRE - Duration::from_secs(1),
});
assert!(
text.trim().is_empty(),
"expired seed must render nothing, got: {text:?}"
);
}
#[test]
fn format_tokens_under_1k() {
assert_eq!(format_tokens_short(0), "0");
assert_eq!(format_tokens_short(1), "1");
assert_eq!(format_tokens_short(10), "10");
assert_eq!(format_tokens_short(100), "100");
assert_eq!(format_tokens_short(999), "999");
}
#[test]
fn format_tokens_1k_to_10k() {
assert_eq!(format_tokens_short(1000), "1.00k");
assert_eq!(format_tokens_short(1230), "1.23k");
assert_eq!(format_tokens_short(1500), "1.50k");
assert_eq!(format_tokens_short(9990), "9.99k");
assert_eq!(format_tokens_short(9999), "10.00k"); // rounds up
}
#[test]
fn format_tokens_10k_to_100k() {
assert_eq!(format_tokens_short(10000), "10.0k");
assert_eq!(format_tokens_short(10100), "10.1k");
assert_eq!(format_tokens_short(12345), "12.3k");
assert_eq!(format_tokens_short(99999), "100.0k"); // rounds up
}
#[test]
fn format_tokens_100k_to_1m() {
assert_eq!(format_tokens_short(100000), "100k");
assert_eq!(format_tokens_short(128000), "128k");
assert_eq!(format_tokens_short(500000), "500k");
assert_eq!(format_tokens_short(999000), "999k");
}
#[test]
fn format_tokens_millions() {
assert_eq!(format_tokens_short(1_000_000), "1.00m");
assert_eq!(format_tokens_short(1_230_000), "1.23m");
assert_eq!(format_tokens_short(9_999_000), "10.00m"); // rounds
assert_eq!(format_tokens_short(10_000_000), "10.0m");
assert_eq!(format_tokens_short(10_100_000), "10.1m");
}
#[test]
fn user_waiting_pulse_speed_is_stable() {
// The drain-blocked, pending-user-input, and plan-approval cues
// all read from this single constant via `pending_diamond_color`,
// so this assertion guards against an accidental tweak that
// would silently change the cadence of every "your turn" cue.
assert_eq!(USER_WAITING_PULSE_SPEED, 0.08);
}
}