You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on Aug 6, 2026. It is now read-only.
-**Claude Code look & feel** — same tool names, calling conventions, and UI patterns (`Agent`, `get_subagent_result`, `steer_subagent`) — feels native
14
+
-**Explicit lifecycle contract** — rejected invocations (invalid models/options/IDs or foreground pre-session startup failure) are Pi tool errors and create no recoverable agent. Once a run is accepted, provider/runtime failures, turn-limit endings, and stops remain successful tool responses with a stable `Agent outcome:` block (`status`, `category`, `agent_id`, and recovery policy). `resume_same_agent` is advertised only for runs that have a session; stopped no-session runs forbid a fresh spawn without inventing recovery, while delayed/background pre-session startup failures retain their accepted ID and allow a fresh start only after correcting the startup problem.
14
15
-**Parallel background agents** — spawn multiple agents that run concurrently with automatic queuing (configurable concurrency limit, default 4) and smart group join (consolidated notifications)
15
16
-**Live widget UI** — persistent above-editor widget with animated spinners, live tool activity, token counts, and colored status icons. Configurable via `/agents → Settings → Widget`: `all` (every agent), `background` (default — hides foreground runs, which already render inline as the `Agent` tool result), or `off`
16
17
-**FleetView** — Claude Code-style navigable list of `main` + every running subagent rendered below the editor (earliest-launched first). Press `↓` (or `←`) at an empty prompt to jump in, `↑`/`↓` to move the selection, `Enter` to open the selected agent's live, auto-updating conversation, `Esc` to return. Finished agents linger briefly before dropping out, and a viewer stays open through completion so you can read the final output. Toggle via `/agents → Settings → Fleet view`
@@ -595,7 +596,7 @@ The agent gets a full, isolated copy of the repository. On completion:
595
596
596
597
The automatic preservation commit uses `--no-verify`, so local pre-commit hooks can't block it — the commit is local-only and never pushed, and pre-push/server-side hooks still apply.
597
598
598
-
If the worktree cannot be created (not a git repo, no commits, or `git worktree add` fails), the `Agent` tool returns a clear error instead of running unisolated — `isolation: "worktree"` is a strict guarantee, not a hint. Initialize git and commit at least once, or omit `isolation`.
599
+
If the worktree cannot be created (not a git repo, no commits, or `git worktree add` fails), the `Agent` tool returns a clear error instead of running unisolated — `isolation: "worktree"` is a strict guarantee, not a hint. Omit `isolation` when an unisolated run is appropriate; never initialize or commit a repository solely to enable worktree isolation.
Copy file name to clipboardExpand all lines: examples/agent-tool-description.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -26,7 +26,7 @@ If the target is already known, use a direct tool — `read` for a known path, `
26
26
- Use model to specify a different model (as "provider/modelId", or fuzzy e.g. "haiku", "sonnet").
27
27
- Use thinking to control extended thinking level.
28
28
- Use inherit_context if the agent needs the parent conversation history.
29
-
- Use isolation: "worktree" to run the agent in an isolated git worktree (safe parallel file modifications). The worktree is automatically cleaned up if the agent makes no changes; otherwise the path and branch are returned in the result.{{scheduleGuideline}}
29
+
- Use isolation: "worktree" only in an existing Git repository with a valid HEAD/at least one commit (safe parallel file modifications). Isolation uses the parent session cwd; a repository path mentioned only in the prompt cannot select another worktree base. Omit isolation for read-only work or a non-Git cwd. Never initialize or commit a repository solely to enable worktree isolation. The worktree is automatically cleaned up if the agent makes no changes; otherwise the path and branch are returned in the result.{{scheduleGuideline}}
0 commit comments