xiaoxue_admin
/
orca
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

refine orca-cli and orchestration skill frontmatter (#1642) 

Sharpens the routing boundary so agents reliably pick the right skill:
orchestration owns agent-to-agent messaging (writes that target another
AI agent's terminal), while orca-cli owns worktrees, browser, and
non-agent terminal I/O — including reads and waits on agent terminals.
Validated against adversarial routing tests.

Co-authored-by: Orca <help@stably.ai>
This commit is contained in:
Jinjing 2026-05-09 18:30:23 -07:00 committed by GitHub
parent 3b87505a97
commit cfa99fd71d
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
skills/orca-cli/SKILL.md
View File

@ -1,6 +1,6 @@
---
name: orca-cli
description: Use the Orca CLI to orchestrate worktrees, live terminals, and browser automation through a running Orca editor. Use when an agent needs to create, inspect, update, or remove Orca worktrees; inspect repo state known to Orca; read, send to, wait on, or stop Orca-managed terminals; or automate the built-in browser (navigate, snapshot, click, fill, screenshot). Coding agents should also keep the current worktree comment updated with the latest meaningful work-in-progress checkpoint whenever useful. Triggers include "use orca cli", "manage Orca worktrees", "read Orca terminal", "reply to Claude Code in Orca", "create a worktree in Orca", "update Orca worktree comment", "click on", "fill the form", "take a screenshot", "navigate to", "interact with the page", "snapshot the page", or any task where the agent should operate through Orca.
description: Use the `orca` CLI to drive a running Orca editor — manage Orca worktrees; create, read, and run shell commands in Orca-managed terminals; and automate Orca's built-in browser (snapshot/click/fill/screenshot/tabs). Use this instead of raw `git worktree`, ad hoc shell PTYs, or Playwright whenever the task touches Orca state. Coding agents inside an Orca worktree should also use it to keep the worktree comment fresh at meaningful checkpoints. Boundary with `orchestration`: if the recipient of a terminal write is another AI agent (Claude Code, Gemini, Codex, a worker), use `orchestration` — it is the only correct way to send messages, nudges, replies, or task hand-offs to agents. orca-cli writes are for non-agent terminals (shells, build/test commands); reading or `wait`ing on any terminal — including agent terminals — stays in orca-cli.
---
# Orca CLI
@ -13,7 +13,7 @@ Use `orca` for:
- worktree orchestration inside a running Orca app
- updating the current worktree comment with meaningful progress checkpoints
- reading and replying to Orca-managed terminals
- reading Orca-managed terminals and sending input to non-agent terminals
- stopping or waiting on Orca-managed terminals
- accessing repos known to Orca
Do not use `orca` when plain shell tools are simpler and Orca state does not matter.
@ -22,7 +22,7 @@ Examples:
- creating one Orca worktree per GitHub issue
- updating the current worktree comment after a significant checkpoint, such as reproducing a bug, validating a fix, or handing off for review
- finding the Claude Code terminal for a worktree and replying to it
- finding the Claude Code terminal for a worktree and reading its status
- checking which Orca worktrees have live terminal activity
## Preconditions

2
skills/orchestration/SKILL.md
View File

@ -1,6 +1,6 @@
---
name: orchestration
description: Use the Orca CLI to coordinate multiple coding agents via inter-agent messaging, task DAGs, dispatch with preamble injection, decision gates, and coordinator loops. Use when an agent needs to send or check inter-agent messages; create, dispatch, or track orchestration tasks; coordinate multi-agent workflows; or act as a coordinator dispatching work across terminals. Triggers include "orchestrate agents", "dispatch task", "send message to agent", "check inbox", "coordinate agents", "multi-agent", "create task DAG", "worker_done", "escalation", or any task involving inter-agent coordination through Orca.
description: Use `orca orchestration` when coordinating multiple AI agents through Orca: pass context, instructions, questions, findings, reviews, handoffs, or structured results between agents; delegate subtasks; dispatch work; track task DAGs; wait for worker_done/escalation; or run coordinator loops and decision gates. This is the right skill for semantic agent-to-agent communication. Boundary with `orca-cli`: do not use orchestration for ordinary terminal control, shell commands, browser automation, worktree management, or reading/waiting on terminals; use `orca-cli` for those. Triggers include "tell the other agent", "ask <agent>", "hand off to", "pass this to <agent>", "delegate to <agent>", "have <agent> do X", "let <agent> know", "share with the other AI", "coordinate agents", "what did <agent> find", "check orchestration inbox", "reply to <agent>", "dispatch task", "orchestrate agents", "multi-agent", "task DAG", "worker_done", "escalation", "decision gate", and "coordinator loop".
---
# Orca Inter-Agent Orchestration