A sophisticated multi-agent orchestration tool that manages group chat interactions, task assignments, and response logic for AI agent teams.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install collaboration-manager
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 collaboration-manager 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 Collaboration Manager?
Collaboration Manager is an essential framework within Openclaw Skills designed to streamline how multiple AI agents interact in shared environments like Feishu (Lark). It solves the common problem of "noise" and overlapping responses by implementing strict @mention priority and keyword-based routing. By defining specific domains for different agents—such as Juna for general help, Code Master for development, and CEVA for finance—it ensures that only the most relevant expert responds to any given prompt.
Beyond simple filtering, this skill enables complex coordination patterns including parallel and serial collaboration. It allows a lead coordinator agent to break down high-level user requests into subtasks, assign them to specialized collaborators, and track progress through structured status updates and interactive message cards. This transforms a collection of individual bots into a cohesive, goal-oriented AI workforce.
Collaboration Manager Use Cases
- Managing large group chats where multiple specialized agents reside to prevent redundant AI replies.
- Coordinating complex software development projects that require input from both System Engineers and Code Masters.
- Executing multi-step financial audits or market analyses involving task handovers between different expert agents.
- Implementing administrative control over AI agents, such as pausing all discussions or muting specific bots for a set duration.
How Collaboration Manager Works
- The system monitors incoming group chat messages and immediately checks for specific @mentions to determine response eligibility.
- If no specific agent is tagged, the skill performs a keyword analysis to route the message to the most relevant agent (e.g., 'bug' routes to Code Master).
- Once an agent is triggered, it identifies itself as either a Lead Coordinator or a Collaborator based on the task structure.
- The Lead Coordinator generates a Task ID and distributes subtasks to other agents using a JSON-based task schema.
- Agents execute their assigned roles (Parallel or Serial) and provide periodic progress updates via interactive Lark message cards.
- Upon completion, the Lead Coordinator synthesizes the results and provides a final summary to the user.
Collaboration Manager Setup
To deploy this skill, ensure your gateway is configured with the correct Feishu/Lark credentials for each agent. Each agent requires a unique openId for effective @mention targeting.
# Install the skill into your workspace
openclaw skill install collaboration-manager
# Configure your group chat policy and agent IDs in config.json
# Then restart the gateway to apply changes
openclaw gateway restart
# Monitor logs to verify agent recognition
openclaw logs --follow | grep feishu
Collaboration Manager Data Schema & Taxonomy
The skill utilizes a structured task and configuration schema to maintain state across the Openclaw Skills environment:
| Object | Key Fields | Purpose |
|---|---|---|
| Task Schema | taskId, coordinator, subtasks, status |
Tracks the lifecycle of a multi-agent assignment. |
| Agent Config | id, name, openId, keywords |
Defines agent identity and domain expertise. |
| Group Config | chatId, members, groupAllowFrom |
Manages permissions and white-listed chat environments. |
Collaboration Manager Advanced Features
- Dynamic @mention Filtering: Strictly enforces silence for agents not explicitly tagged to ensure channel clarity.
- Collaboration Modes: Support for 'Parallel' (independent modules) and 'Serial' (dependent workflows) task execution.
- Administrative Overrides: Command-line style triggers like "Stop all discussions" or "Mute [Agent] for 5 minutes" for human operators.
- Interactive Message Cards: Real-time visual progress tracking using formatted Lark/Feishu templates.
- Automated Task Synthesis: Automatic generation of completion reports and time-duration tracking for complex workflows.
SKILL.md
Loading
Related Openclaw Skills
MIC Recorder Skill
A high-quality microphone recording skill for macOS that automatically denoises audio and sends it to Feishu.
Excel Parser
A high-performance Python-based utility for parsing and extracting structured data from multiple Excel file formats.
Toutiao Publisher
An automation tool for publishing short-form text and image content to the Micro-Toutiao platform via a browser-based workflow.
API Cost Tracker
A comprehensive utility for tracking, analyzing, and optimizing LLM API expenditures across multiple providers within your agent environment.
ClawHub Publisher
An automated assistant designed to publish local skill directories directly to clawhub.com while handling common CLI bugs and deployment hurdles.
PDF OCR Skill
A dual-engine OCR tool that enables AI agents to extract text from scanned PDF files and various image formats using local or cloud-based AI models.
Featured*
