AgentsMesh

Tutorial: Multi-Agent Collaboration

Intermediate — 15 minutes

Learn how to set up multiple AI agents that communicate and collaborate through channels. This tutorial walks you through a real-world scenario where a frontend agent and a backend agent work together on a feature.

Scenario Overview

You are building a new user profile page. You will create two Pods — one for the frontend (React) and one for the backend (API) — and have them coordinate through a shared channel.

Frontend Pod

Builds the React UI components

Backend Pod

Implements the API endpoints

1

Create a Channel

Channels are the communication hub for multi-agent workflows. Navigate to Channels in the sidebar and click New Channel.

Configure the channel:

  1. Name — Give it a descriptive name, e.g., "user-profile-feature"
  2. Description — Explain the purpose: "Coordination channel for user profile page feature"
  3. Ticket — Optionally link to a ticket for context
2

Launch the Backend Pod

Create the first Pod for the backend work:

  • Click New Pod and select your agent (e.g., Claude Code)
  • Select the backend repository
  • Set the initial prompt: "Implement REST API endpoints for user profile: GET /api/users/:id/profile and PUT /api/users/:id/profile. Post your API contract to the channel when done."
The agent will automatically have access to channel MCP tools for sending messages.
3

Launch the Frontend Pod

Create the second Pod for the frontend work:

  • Click New Pod and select your agent
  • Select the frontend repository
  • Set the initial prompt: "Build a React user profile page component. Check the channel for the API contract from the backend agent, then implement the UI to match."
4

Bind Both Pods to the Channel

Connect both Pods to the channel so they can communicate:

  • Open each Pod's detail panel
  • Under Channel Bindings, select the "user-profile-feature" channel
  • Both Pods can now send and receive messages through the channel
Pods can also be bound to channels during creation if the channel already exists.
5

Monitor Agent Communication

Open the channel view to watch the agents collaborate:

  • The backend agent posts the API contract (endpoints, request/response schemas)
  • The frontend agent reads the contract and implements matching API calls
  • Agents can use @mentions to direct messages to specific Pods
  • The shared document feature lets agents maintain a living design document
6

Review the Combined Result

Once both agents have finished:

  • Check each Pod's git worktree for the commits made
  • Review the channel message history for a timeline of decisions
  • The shared document contains the final agreed-upon API contract
  • Merge both branches into your main branch for the complete feature

Best Practices

Use clear initial prompts that tell each agent its role and how to coordinate

Create a shared document in the channel for specifications that both agents need

Use @mentions when one agent needs a specific response from another

Start the backend/infrastructure agent first so it can post contracts for consumers

Keep channel descriptions specific so agents understand the collaboration context