TeamAI CLI

// Overview

TeamAI CLI is an open-source command-line tool from Tencent that treats a git repository as the source of truth for how a team's AI coding agents behave. An admin publishes skills, rules, docs, environment variables, hooks, agents and MCP servers into that repo; every member runs `teamai init` against its URL once, and from then on each AI session pulls whatever the admins have published, with no manual sync step.

The tool is organised in three layers. Team Execution is the part that is fully shipping: `init`, `pull` and `push` for the shared harness of skills, rules, agents, hooks, MCP and env. Team Context (beta) adds recall, shared learnings, a codebase graph and a team wiki so an agent starts a session knowing something about the codebase. Team Improvement (beta) collects session friction, usage stats and a digest so real sessions turn into new shared skills and rules.

Support is deliberately multi-agent. The README's coverage table gives Claude Code, Codex, Cursor, CodeBuddy and Qoder a tick in every column, with narrower support for WorkBuddy, OpenCode, OpenClaw, Hermes, DeepSeek Harness and ZCode. Resources install at project scope by default or at user scope with `--scope user`. TeamAI CLI is published on npm as `teamai-cli` under the MIT license and needs Node.js 20 or newer plus git.

// What it does

• One shared git repo holds the team's skills, rules, docs, env, hooks, agents and MCP servers; members pull the latest automatically on every AI session
• Multi-agent by design: Claude Code, Codex, Cursor, CodeBuddy, WorkBuddy, OpenCode, OpenClaw, Hermes, DeepSeek Harness, Qoder and ZCode
• `teamai push` sends a local skill or rule back to the team repo through a merge-request style review flow
• Project scope by default, or `--scope user` to install shared resources under the home directory for every project on the machine
• Team Context (beta): recall, shared learnings, a codebase graph and a team wiki so agents do not start each session from zero
• Team Improvement (beta): session records, usage stats and a digest dashboard built from friction signals in real sessions
• Works with GitHub, GitLab (hosted or self-hosted), GitCode, CNB, TGit and private git services

// Getting started

TeamAI CLI installs from npm and needs Node.js 20 or newer plus git. One admin creates the shared team repo; everyone else only runs `teamai init` against its URL.

npm install -g teamai-cli
teamai --version

teamai init https://github.com/yourorg/yourrepo

cd /path/to/my-project
teamai init https://github.com/yourorg/yourrepo

# or, for every project on this machine
teamai init https://github.com/yourorg/yourrepo --scope user

export GITLAB_URL=https://git.example.com
export GITLAB_TOKEN=glpat-xxxxxxxxxxxxxxxx
teamai init https://git.example.com/yourgroup/yourrepo

teamai packages

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Stop copy-pasting the same agent rules into a dozen private config directories and let a reviewed git workflow distribute them instead
• Give a new joiner a working agent configuration on day one by having them run a single `teamai init`
• Keep a shared setup working when team members use different agents, so someone moving from Cursor to Claude Code keeps the same skills and rules
• Publish a skill you wrote locally back to the whole team through `teamai push` and a merge request

// How TeamAI CLI compares

TeamAI CLI alongside other open-source agent skills & plugins tools AI/TLDR tracks, ranked by GitHub stars.