PlugWiki

orchestration Skill for Claude: Run Orca Worker Agents

orchestration lets your agent hand work to separate Orca worker agents, such as Codex from Claude, and get the results back.

Installs
353.9K
on skills.sh
Stars
112.4K
on GitHub
Updated
Nov 2026
last repo commit

orchestration is a launcher skill for Orca. It triggers when you ask for Orca workers or supervision, such as monitoring, task graphs, ask and reply, or decision gates. It also triggers when the work needs a different agent or its own Orca workspace, for example "have Codex review this and report back".

For other work it tells the agent to use its own subagents. It needs the Orca app running, because the full guide comes from orca skills get orchestration and the skill says never to swap in a non-Orca subagent tool when you asked for Orca.

Good for

  • Developers who want a second agent to review or build something and report back
  • Orca users who supervise several agents with task graphs and decision gates
  • Agents dispatched as Orca workers that need to ask questions or report done

Install orchestration

Claude Code

  1. 1

    Copy this prompt.

    Install the orchestration skill in Claude Code. Run this command:

npx skills add https://github.com/stablyai/orca --skill orchestration

Then check that it loaded and tell me how to use it.
  2. 2

    Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

From its listing on skills.sh. Source.

What's in the skill

When it loads

Run work in separate Orca worker agents and get their results back. Use when the user asks for Orca workers or supervision (monitor, task DAGs, ask/reply, decision gates), or when the work needs something your own subagents cannot give: a different agent (such as Codex from Claude) or its own Orca workspace. "Have Codex (or Claude) review this and report back" means an Orca worker of that agent, even when you have your own model subagents. Also use when you are a dispatched worker (a prompt with Task and Dispatch IDs: ask, worker_done). Otherwise prefer your own subagents, including for long or parallel work. Use `orca-cli` to hand work off with nothing expected back ("hand off", "handover", "give this to another agent", "another worktree"), and for terminal control, shell commands, worktree management, and reading or waiting on terminals.

Sections in SKILL.md

  1. Resolve the CLI for this session
  2. Load the version-matched guide before running Orca commands

Key facts

Author
stablyai (owner of the stablyai/orca repo)
License
MIT
Installs on skills.sh
353,917 as of 2026-11-02
GitHub stars
112,427 for the whole stablyai/orca repo
skills.sh audits
Gen Agent Trust Hub warn, Socket pass, Snyk pass
What is inside
SKILL.md only; bundles no other files
Needs
The Orca desktop app and its command line tool; no API key is named in SKILL.md
Last commit to the repo
2026-11-01

What it includes

  • Chooses which Orca command to run, once per session

    Uses ORCA_CLI_COMMAND if set, then orca-dev in a dev checkout, then orca-ide on Linux outside an Orca terminal, otherwise orca.

    Read · Pick the Orca executable

  • Loads the guide for the local coordinator loop

    Prints the guide that matches your Orca version. It covers the coordinator loop and what a dispatched worker must do.

    Try: “Have Codex review this branch in its own Orca workspace and report back.”

    Runs code · orca skills get orchestration

  • Reads extra guides only when a gate needs them

    Gates such as remote placement, uncertain release recovery or expanded task graphs load one file with --reference. --references lists the names. If the binary rejects --reference, SKILL.md says to run --full and read the named reference.

    Runs code · Load a reference on demand

  • Lets a dispatched worker ask questions and report

    A worker started with Task and Dispatch IDs sends ask questions and a worker_done message back to the coordinator.

    Runs code · Worker messages: ask and worker_done

  • Starts Orca if a command says it is not running

    Starts the app, then retries the command.

    Runs code · orca open --json

From the official docs

Permissions

Medium risk

The skill starts other agents through Orca. Check what each agent is allowed to do before you let a worker start.

What it can do

  • Read your data
  • Create and change things
  • Delete things
  • Read-only mode

Watch out for

  • Work runs in separate worker agents

    SKILL.md has the coordinator hand tasks to other agents such as Codex, each in its own Orca workspace. It does not describe their permissions, so check each agent's own settings before a worker starts.

  • It may ask for escalated permissions

    If your sandbox blocks the connection to Orca, SKILL.md has the agent re-run the command with escalated permissions. Approve that only for the Orca command.

Technical details
Scopes

  • Runs the orca command on your machine

  • Starts and supervises other agents in separate Orca workspaces

Questions

What does the orchestration skill do?

It makes your agent run work in separate Orca worker agents and collect their results. Typical uses are a review by Codex while you work in Claude, or a task that needs its own Orca workspace.

When does it trigger instead of normal subagents?

Only when you ask for Orca workers or supervision, or the work needs a different agent or its own Orca workspace. Otherwise SKILL.md tells the agent to use its own subagents, even for long or parallel work.

How do I install orchestration?

Use the install box at the top of this page, which has tabs for Claude Code, the Claude app and Codex. Orca itself must be installed and running.

Does it need an API key?

No key is named in SKILL.md. Whether a worker agent such as Codex or Claude needs its own sign-in is not covered there.

orchestration or orca-cli?

Use orchestration when a worker should do a task and send back a result. Use the orca-cli skill to hand work off with nothing expected back, or for terminals, worktrees and the browser. They are separate skills, so install both if you want both.

Is orchestration safe?

It starts other agents in separate Orca workspaces, and SKILL.md does not describe what those agents may do, so check each agent's own permission settings first. The skill folder holds only SKILL.md. See the safety section for details.

Information

updated Nov 2, 2026

Repository
stablyai/orca
Folder
skills/orchestration

stablyai/orcaskills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 3
  1. orchestration SKILL.md · github.com
  2. stablyai/orca README · github.com
  3. skills.sh listing and audits · skills.sh

Updated Nov 2, 2026.