Overview
TeamAI is Tencent Cloud open-source team AI collaboration tool. It puts the way a team uses AI into a Git repository: skills, working standards, tool configuration and project knowledge are reviewed, updated and versioned like code, then synced to whatever agent each member runs.
Key Features
- Git as the source of truth: Skills, rules, docs, sub-agents, hooks, MCP configuration and environment variables all live in a shared repository, with changes going through branches and merge requests
- Works with many agents: Supports 16 agents including WorkBuddy, CodeBuddy, Claude Code, Codex and Cursor, plus six Git providers from GitHub and GitLab to GitCode, CNB, TGit and private Git services
- Automatic sync at session start: After a merge, agents with a session-start hook pull the latest configuration when a new conversation opens, with no per-person notifications or manual replacement
- Experience captured automatically: When repeated corrections or failed tool calls show up, it prompts the member to share, and the agent writes it up as a document that later tasks can retrieve
- Codebase knowledge graph: Maps references and call relationships across code so agents can locate the modules involved and cite source locations for verification
Use Cases
- Distributing skills and standards team-wide, so new members start with the team accumulated configuration
- Letting an admin set model services that meet security requirements, lowering the risk of members picking the wrong model
- Turning recurring troubleshooting steps into documents that later tasks can search
- Sharing one base across sub-brands or project groups, distributing different configuration by role, project or tag
Pros
- Manages AI assets with the Git workflow teams already know, so review and rollback use familiar tooling
- One configuration syncs to many agents, so members need not abandon the tool they prefer
- The agent writes up captured experience, reducing the burden of documentation
- In official testing, a lower-cost model with team knowledge scored 6.1% higher overall while model call costs fell 76%
Pricing
An open-source project under the MIT license, hosted at Tencent/teamai-cli on GitHub and installed globally via npm; costs for whichever model service you configure are your own.
Summary
TeamAI fits engineering teams of three or more whose members use different agents and who want AI experience to accumulate at the team level. Its value grows with team size; for a small team or an individual, the overhead of the Git review flow outweighs the benefit.
Version History
- Tencent Cloud open-sources TeamAI (2026-10-10): Tencent Cloud open-sourced TeamAI, the team AI collaboration tool it had been using internally. A Git repository acts as the source of truth for team AI assets: submitting a skill or changing a standard automatically opens a branch and a merge request, and once merged, agents with a session-start hook pull the update when a new conversation begins. Sharing covers MCP tool configuration, rules and sub-agents, distributable by role, project or tag, with centralized model configuration distribution newly added. In an internal test across 13 engineering tasks, a lower-cost model with team knowledge scored 6.1% higher overall while model call costs fell 76%.