Feishu Agent Relay

What This Skill Does

The Feishu Agent Relay by glassmarbles is an advanced orchestration framework designed to build multi-agent systems directly within the Feishu (Lark) ecosystem. It allows developers to create a centralized 'Coordinator' bot that intelligently routes tasks to specialized 'Expert' bots, enabling seamless collaboration. The skill handles complex architectural patterns, including cross-agent identity mapping, message relaying, and proactive agent messaging, ensuring that users can interact with a single interface while leveraging the power of multiple specialized AI agents in the background.

Installation

Installation is streamlined through the ClawHub system. First, determine your deployment strategy. Use the command clawhub install openclaw/skills/skills/glassmarbles/feishu-agent-relay to initialize the library. If choosing Single-User mode, simply set DEPLOYMENT_MODE=single-user in your environment to bypass manual configuration and identity mapping. For enterprise environments, follow the references/feishu-bot-setup.md guide to provision multiple Feishu bot applications and configure your relay table accordingly.

Use Cases

• Corporate Customer Support: Create a 'Front Desk' bot that routes queries to either technical support, billing, or sales specialist agents based on natural language analysis.
• Project Management: Build a multi-bot system where a 'Project Coordinator' manages tasks while separate 'Developer' and 'Designer' bots handle technical execution and feedback collection.
• Proactive Notifications: Enable specialized agents to send updates directly to users via Feishu when specific milestones are reached or tasks require immediate attention.

Example Prompts

1. "Check with the Technical Support agent if there is any update on my open ticket #4092."
2. "I need to speak with the Billing expert; please route my request for an invoice correction."
3. "Summarize the status of all active tasks assigned to the Designer bot for this week."

Tips & Limitations

This skill is highly powerful but requires careful planning of bot infrastructure. Always prefer Single-User mode for personal projects to avoid the security risks associated with manual identity registration. Note that Multi-User mode currently lacks built-in identity verification, so it should be restricted to internal, trusted environments. Ensure that each specialist agent is configured with clear scope boundaries to prevent role overlap during coordination tasks.