team-chat

File-backed team collaboration protocol for AI agents.

Repo: fractalmind-ai/team-chat-skillLanguage: Python Status: Stable Install: npx openskills install fractalmind-ai/team-chat-skill

What It Does

team-chat provides asynchronous, file-backed messaging between AI agents:

• Append-only inboxes: Each agent has a JSONL inbox file
• Event logs: Daily event streams for audit
• Task state snapshots: Track task assignments and updates
• Acknowledgements: Agents confirm receipt of messages
• Retry/dead-letter: Failed messages get retried with safety guarantees

Key Design

• File-first: All messages stored as JSONL files — no database, no server
• Append-only: Messages are never deleted or modified
• Atomic writes: Safe for concurrent access
• Observable: status and trace commands for debugging

Usage

npx openskills read team-chat

# Send a message to an agent
team-chat send --to EMP_0003 --message "Please review the SDK changes"

# Read inbox
team-chat read --agent EMP_0001

# Acknowledge a message
team-chat ack --message-id msg_123

# Assign a task
team-chat task_assign --to EMP_0003 --task "Fix failing tests in CI"

# Check status
team-chat status

Where It Fits

team-chat handles agent ↔ agent communication:

fractalbot  →  human ↔ agent  (external)
team-chat   →  agent ↔ agent  (internal)

It is the async messaging backbone for team-manager coordination. When a lead assigns work, the assignment goes through team-chat. When a member completes work, the result is posted to team-chat.

Related Components

• team-manager — uses team-chat for coordination
• fractalbot — handles external (human) communication
• agent-manager — manages the agents that use team-chat