A professional-grade orchestration protocol for managing multi-agent teams with automated task tracking, approval gates, and asynchronous dispatch.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-orchestration-kit
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install agent-orchestration-kit using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Agent Orchestration Kit?
The Agent Orchestration Kit is a specialized framework designed to transform individual AI agents into a high-performance team. Unlike simple prompt packs, this kit implements a rigorous orchestration protocol using Openclaw Skills to govern complex interactions. It establishes a star topology where a central Leader agent manages specialist spokes, ensuring that work is decomposed, dispatched, and tracked with absolute precision.
By leveraging this kit, developers can build resilient AI workflows that feature built-in safety nets, such as heartbeat-based stale task detection and mandatory approval gates. It provides the structural integrity needed for scaling AI operations, moving beyond basic chat interactions into a structured environment where every task is a traceable file and every agent follows a standardized callback protocol.
Agent Orchestration Kit Use Cases
- Coordinating multi-agent teams for complex software development or system architecture projects.
- Managing high-volume content production pipelines using specialized researcher and creator agents.
- Implementing secure AI workflows that require human-in-the-loop approval before external execution.
- Scaling asynchronous AI tasks where the main agent must remain unblocked while specialists work.
- Automating long-running research tasks with built-in recovery for context loss or session timeouts.
How Agent Orchestration Kit Works
- The Leader agent receives an owner request and decomposes it into atomic, actionable tasks.
- A dedicated task file is created in the tasks directory to serve as the single source of truth before any work begins.
- Tasks are dispatched asynchronously to specialist agents (spokes) using non-blocking session protocols.
- Specialists perform the work and return structured signals such as [READY], [BLOCKED], or [NEEDS_INFO].
- The Leader processes callbacks, updates the task file status, and manages quality review cycles.
- Upon completion, the result enters an approval pipeline, holding for owner confirmation before final execution or delivery.
Agent Orchestration Kit Setup
Ensure you have OpenClaw v2026.2.26+ installed and your auth profiles configured. Follow these steps to deploy Openclaw Skills for orchestration:
- Run the scaffold script to generate your team structure:
bash scripts/scaffold.sh --skill-dir "$(pwd)" --team <selected-team>
- Preview the configuration changes to ensure they align with your environment:
node scripts/patch-config.js --config ~/.openclaw/openclaw.json --dry-run
- Apply the configuration patch to enable agent-to-agent communication and heartbeat monitors:
node scripts/patch-config.js --config ~/.openclaw/openclaw.json
- Restart the gateway and verify your setup:
openclaw gateway restart
openclaw doctor
Agent Orchestration Kit Data Schema & Taxonomy
The kit organizes data within the ~/.openclaw/workspace/ directory using a transparent file-based taxonomy:
| Directory/File | Description | Purpose |
|---|---|---|
tasks/ |
Markdown task files | Tracks live status, steps, and logs for every task. |
shared/INSTANCE.md |
Instance Metadata | Stores owner name, timezone, and language preferences. |
shared/operations/ |
Protocol Rules | Contains team rosters, signal definitions, and brief templates. |
HEARTBEAT.md |
Health Check | Defines the procedure for the 3-minute automated status check. |
SOUL.md / AGENTS.md |
Agent Logic | Defines the persona and operating procedures for each agent. |
Agent Orchestration Kit Advanced Features
- Heartbeat Safety Net: Automatically detects and recovers stale tasks every 3 minutes.
- On-Demand Reviewers: Dynamically spawns reviewer agents for high-stakes deliverables or repeated failures.
- Context Loss Recovery: Uses the task file's source of truth to re-brief agents if a session is compacted.
- External Agent Integration: Compatible with over 100+ templates from the agency-agents catalog.
- Customizable Approval Pipeline: Define specific phrases for shortcuts and set custom stale-approval timeouts.
SKILL.md
Loading
Related Openclaw Skills
Ads Optimizer
Ads Optimizer provides AI-driven campaign management including performance audits, optimization recommendations, and automated anomaly detection.
CRM Engine
A complete CRM system built for autonomous AI agents to manage accounts, leads, deals, and work orders via a persistent API layer.
After School Program Video
A specialized video production skill designed to transform childcare credentials and developmental data into high-converting parent acquisition videos.
AI Online Course Empire
A comprehensive AI-driven workflow that transforms niche expertise into a launch-ready online course with validated market data and high-conversion marketing assets.
Scheduling Engine
A programmatic scheduling engine designed for AI agents to manage staff availability, detect conflicts, and automate appointment booking.
Power User Pack
A premium collection of four essential tools designed to optimize memory, automate git workflows, and stabilize agent performance within the Openclaw Skills ecosystem.
Featured*
