TeamAI CLI

TeamAI CLI is an open-source command-line tool published by Tencent under the MIT license that helps engineering teams standardize and distribute AI agent configurations across their entire organization. It works by syncing skills, rules, docs, hooks, MCP settings, and environment variables from a shared Git repository to every team member's local AI tools — supporting Claude Code, Codex, Cursor, CodeBuddy, WorkBuddy, OpenCode, and more. The project has accumulated over 4,000 GitHub stars and 269 forks since its creation.

What It Is

TeamAI CLI is a team-level AI harness manager: it treats your team's AI agent configuration as code, stored in a shared Git repo and distributed to every developer through a push → review & merge → pull workflow. Rather than each developer independently configuring their AI tools, a team admin defines skills, rules, agents, hooks, and MCP settings once, and every member receives them automatically on teamai pull. The tool is installed globally via npm (npm install -g teamai-cli) and works with any Git provider including GitHub, GitLab, GitCode, CNB, TGit, and private Git services.

Three-Layer Architecture

The README describes three product layers, two of which are currently in beta:

• Team Execution — The stable core: init, pull, push, skills, rules, agents, hooks, MCP, and env distribution across all supported agents.
• Team Context (beta) — Organizes accumulated team experience and code structure into a searchable knowledge base. Includes automatic experience sharing triggered by session "friction" (AI corrections, denied tool calls, retries), a BM25 + graph-boosted recall system, and a codebase knowledge graph built from AST parsing (TypeScript, JavaScript, Python, Go via WASM tree-sitter) and heuristic extraction for all other languages.
• Team Improvement (beta) — Turns session data into team intelligence: weekly usage digests, privacy-scrubbed session summaries, a web dashboard with live session views, and a knowledge-base health console.

Distribution Controls and Filtering

Admins configure team-wide settings once and deliver them to every member on teamai pull. The tool supports fine-grained filtering so members only receive what's relevant to them:

• Projects — Bind a working directory to one or more logical projects for isolated skills and learnings.
• Roles — Define role-to-namespace mappings so each member syncs only the skills for their role.
• Tags — Tag skills and rules so members subscribe to just the tags they need.
• Sources — Subscribe to additional skill repos from other teams or shared org repos; subscribed skills sync automatically on pull.

Supported Agents and Coverage

The README provides a detailed compatibility matrix. Claude Code, Codex, Cursor, CodeBuddy, and Qoder have full support across all thirteen capability columns (skills, rules, docs, env, agents, hooks, MCP, learnings, codebase, teamwiki, usage, sessions, dashboard). WorkBuddy, OpenCode, ZCode, OpenClaw, Hermes, and DeepSeek Harness have partial support depending on the capability. This breadth makes TeamAI CLI agent-agnostic at the team level.

Update: v0.23.1

The latest release is v0.23.1, published on September 9, 2026, with the repository last pushed on September 11, 2026. The project was created in April 2026 and has seen rapid iteration, reaching over 4,100 stars in its first months. The primary language is TypeScript. The teamai-hub GitHub organization provides template repositories pre-loaded with production-ready skills, rules, and review agents to help teams get started quickly.