本文へ移動
cccskills
無料GitHub で公開

ao-orchestrator

Use when acting as the multi-project orchestrator and lead coordinator across repositories using the Agent Orchestrator (AO) daemon. Spawns, inspects, coordinates, and reviews parallel coding agents in isolated Git worktrees.

インストール方法を見る

含まれるファイル(1)

  • SKILL.md4.7 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

👑 AO Multi-Project Orchestrator (/ao-orchestrator)

Overview

This skill equips any AI coding agent across all supported harnesses (Claude Code, Google Antigravity, OpenCode, Cursor, Codex) to act as a Multi-Project Orchestrator supervising tasks across independent code repositories via the Agent Orchestrator (AO) background daemon (~/.local/bin/ao).

Instead of making changes directly across multiple repositories within a single context window, the Orchestrator breaks down high-level business goals into scoped work items, delegates them to specialized worker agents running in isolated Git worktrees, monitors their progress, and reviews/integrates their Pull Requests.


🧭 Orchestrator Responsibilities

  1. Project Discovery & Mapping: Identify which repositories and workspaces are involved in the user's goal.
  2. Task Decomposition: Split cross-repo features into clear, atomic specifications for individual worker sessions.
  3. Session Lifecycle Management: Spawn worker agents with explicit roles (architect, techlead, ui_ux, engineering, reviewer, shipping), appropriate harnesses (claude-code, opencode, agy), and specific prompts.
  4. Coordination & Supervision: Track running sessions via ao session ls, inspect agent logs, and inject steering instructions or clarifications via ao send.
  5. Quality Gate & PR Review: Run ao review once a worker completes its task, verify tests, and coordinate cross-project integration.

🛠️ AO CLI Core Command Suite

The Orchestrator interacts with the local daemon (listening on loopback port 3101) strictly through the ao CLI:

1. Discovery & Status

# Check AO daemon health and version
ao status

# List all registered repositories and workspaces
ao project ls

# List active and recent agent sessions across all projects
ao session ls

2. Spawning Multi-Project Workers

Run worker agents in fresh, isolated Git worktrees (ao/<session-id>/root) so they never conflict with uncommitted working trees:

# Spawn a Backend Engineer in Project A
ao spawn \
  --project <project-id-or-path> \
  --name "Backend-API" \
  --role engineering \
  --harness claude-code \
  --prompt "Implement the OAuth2 callback handler per specification in issue #42. Ensure all tests pass."

# Spawn a Frontend Designer/Engineer in Project B
ao spawn \
  --project <project-id-or-path> \
  --name "Frontend-UI" \
  --role ui_ux \
  --harness claude-code \
  --prompt "Build the settings page OAuth connection cards with Tailwind and React."

# Spawn an Adversarial Reviewer for an existing worker branch
ao spawn \
  --project <project-id-or-path> \
  --name "Audit-Worker" \
  --role reviewer \
  --prompt "Review branch ao/<session-id>/root for security gaps, unhandled errors, and contract compliance."

3. Monitoring & Steering Sessions

# Inspect detailed state of a specific session
ao session get <session-id>

# Send instructions, feedback, or corrections to a running worker
ao send <session-id> "The database migration must support SQLite and Postgres. Please update the schema."

# Switch agent harness if needed (e.g. switch to opencode or agy)
ao session switch-agent <session-id> --to opencode

4. Review & Completion

# Trigger an AO code review of the worker's changes
ao review <session-id>

# Clean up or terminate completed sessions
ao session kill <session-id>
ao session cleanup

📋 Multi-Project Orchestration Protocol

When the user asks you to coordinate work across multiple projects:

  1. Step 1: Check Registered Projects Run ao project ls to verify the repositories exist in AO. If a project is not registered, instruct the user to register it or navigate to the directory and run ao project add ..
  2. Step 2: Propose the Architecture & Work Breakdown Present a clear plan to the user:
    • Which repo needs what changes.
    • What workers will be spawned and with which AOS roles.
    • The dependency order (e.g. Backend API first, then Frontend integration).
  3. Step 3: Spawn Workers Orderly Spawn the prerequisite workers. Avoid spawning dependent workers before the contract/API they depend on is established.
  4. Step 4: Report Progress to User Provide regular, concise summaries in your response showing the status of each spawned session and next actions.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Use when manage multiple local CLI agents via tmux sessions (start/stop/monitor/assign) with cron-friendly scheduling.

日本語の概要は準備中です。原文の説明を表示しています。

hybridlabor-api/aos62026年10月12日 更新

Use when a hybrid memory system that provides persistent, searchable knowledge management for AI agents (Architecture, Patterns, Decisions).

日本語の概要は準備中です。原文の説明を表示しています。

hybridlabor-api/aos62026年10月12日 更新

Reference for how BDB structures autonomous software engineering work — the seven-node dispatcher graph (Architect, TechLead, UI/UX, Engineering, Media/EventTech, Reviewer, Shipping) that /startcycle-graph actually runs. Use when you need the high-level lifecycle framing without inventing your own process.

日本語の概要は準備中です。原文の説明を表示しています。

hybridlabor-api/aos62026年10月12日 更新

Tools are how AI agents interact with the world. A well-designed tool is the difference between an agent that works and one that hallucinates, fails silently, or costs 10x more tokens than necessary. This skill covers tool design from schema to error handling.

日本語の概要は準備中です。原文の説明を表示しています。

hybridlabor-api/aos62026年10月12日 更新

Harness patterns for coding agents — memory, permissions, context engineering, delegation, skills, hooks, bootstrap.

日本語の概要は準備中です。原文の説明を表示しています。

hybridlabor-api/aos62026年10月12日 更新

agenttrail: live map of a multi-agent build in the browser: which plan component is being worked on, by which agent or harness, what is done and what is stuck. Use when a multi-agent pipeline starts (/startcycle, /startcycle-graph, /teamwork-preview) or after a plan-canvas approve, when the user asks to see what the agents are doing ("live map", "build board", "who is running now"), or by yourself whenever a multi-agent run is under way — no user prompt needed.

日本語の概要は準備中です。原文の説明を表示しています。

hybridlabor-api/aos62026年10月12日 更新

hybridlabor-api のスキルをすべて見る

このスキルの問題を報告する