Define your AI team once. Sync it everywhere.

teamrc.ai · Docs

If you keep recreating the same agents, skills, and prompts across Claude Code, Cursor, Codex, Gemini, OpenClaw, and different machines, teamrc gives you one source of truth.

Define your setup once in .teamrc.yaml, then run teamrc sync to write the native files your installed AI tools expect.

• One shared definition for agents, skills, and team knowledge
• Local-first — no server required for solo use
• Optional relay sync for multi-machine and team workflows
• Native output for Claude Code, Cursor, Codex, Gemini, and OpenClaw

npx @teamrc/cli init --local
npx @teamrc/cli sync

No account, no server. Your files stay local unless you later connect a relay.

npx @teamrc/cli join <invite-code>
npx @teamrc/cli sync

After sync, teamrc writes native agent, skill, and config files for the AI tools installed on this machine.

name: my-team

members:
  - name: researcher
    role: Finds and summarizes relevant information
  - name: builder
    role: Implements and iterates on ideas

skills:
  - id: write-tests
    body: Always write tests for new functionality.
    alwaysApply: true

npx @teamrc/cli sync

This writes native config files for each installed tool — agent definitions, skill rules, and shared knowledge — so researcher and builder are available everywhere.

---
config:
  flowchart:
    curve: linear
---
graph TD
    R["Teamrc Relay"] <--> A["Machine A"]
    R <--> B["Machine B"]
    R <--> C["CI / VM"]

    A --> A1["Claude Code\nGemini"]
    B --> B1["Cursor\nOpenClaw"]
    C --> C1["Codex\n..."]

    style R fill:#e0e7ff,stroke:#6366f1,color:#312e81
    style A fill:#f4f4f5,stroke:#a1a1aa,color:#27272a
    style B fill:#f4f4f5,stroke:#a1a1aa,color:#27272a
    style C fill:#f4f4f5,stroke:#a1a1aa,color:#27272a
    style A1 fill:#fff,stroke:#d4d4d8,color:#52525b
    style B1 fill:#fff,stroke:#d4d4d8,color:#52525b
    style C1 fill:#fff,stroke:#d4d4d8,color:#52525b

Each machine runs teamrc sync to write native files for its installed AI tools. If you connect a relay, team config and shared knowledge stay in sync across machines and teammates.

activeMembers controls which agents are installed in the current project. It is local to each machine and project — it does not change the team on the relay.

By default, all team members are active. During init and join, an interactive chooser lets you pick which agents should be active on this machine. You can also pass --members explicitly:

# Interactive: prompts you to select active agents
teamrc init
teamrc join <invite-code>

# Explicit: skip the picker
teamrc join <invite-code> --members frontend,designer
teamrc init --members frontend,designer

# Change active members later
teamrc members

docker compose up   # starts Postgres + relay at http://localhost:4000

Point the CLI at your relay with TEAMRC_RELAY=http://localhost:4000 or set relay: in .teamrc.yaml.

See docs/deploy-coolify.md for production deployment.

• Get Started — Create a synced team in 2 minutes
• Platforms — How teamrc works on each AI platform
• Configuration — .teamrc.yaml format and options

• CLI Reference — All commands with flags and examples
• Sharing & Access — Public sharing, access roles, clone/fork
• Template Catalog — 12 team templates, 68 agents, 49 skills

• Architecture — System design, auth, sync protocol
• API Reference — REST and WebSocket API