teamai-cli - Tencent's open-source AI team collaboration CLI tool

What is teamai-cli

TeamAI CLI is Tencent's open-source MIT-licensed CLI tool for unified management of a team's AI programmingconfiguration. The tool stores resources such as skills, rules, MCP, and hooks in a Git repository and distributes them to each member's local AI tools through a push→MR→merge→pull workflow. TeamAI CLI supports friction-signal-driven knowledge capture, team knowledge recall, and codebase graphs, making team AI collaboration standardized.

Main features of teamai-cli

• Unified configuration management: Centrally manage the team's skills, rules, MCP, hooks, env, and other resources in a single Git repository.
• Git-native distribution: Automatically sync team configuration to each member's local AI tools through a push→MR→merge→pull workflow.
• Multi-Agent compatibility: Simultaneously supports Claude Code、Cursor、Codex、CodeBuddy、WorkBuddy、Qoder and 9+ other AI programming tools.
• Team-level governance: Use Roles, Tags, and Sources to achieve fine-grained subscription and filtering by role, by topic, and across teams.
• Unified team MCP configuration: Declare once in YAML, and automatically convert it into the native MCP configuration files of each tool.
• Team Hooks event system: Inject custom scripts such as security scans and compliance interception at event points like SessionStart and PreToolUse.
• Friction signal knowledge capture: Automatically detect friction signals in sessions such as interruptions, refusals, and retries, trigger experience summaries, and push them to the team repository.
• Team knowledge recall: Automatically search the team knowledge base before a task (BM25 + graph-enhanced ranking), and automatically skip unrelated tasks.
• Codebase knowledge graph: Parse source AST and heuristic dependencies to build a cross-repository structured graph for recall enhancement.
• Team Analytics: Provide weekly digest reports, privacy-stripped session summaries, and a real-time Web Dashboard.

Follow WeChat and reply "open source" to join theAI open-source project exchange group

How to use teamai-cli

• Install the CLI: Install globally npm install -g teamai-cli。
• Create a team repository: Create a shared repository on platforms such as GitHub/GitLab and grant team members write access.
• Initialize team configuration: The administrator runs teamai init <仓库地址> to complete OAuth login, repository association, and member registration.
• Members join the project: Team members enter the project directory and run teamai init <仓库地址>, then choose project-level or user-level installation.
• Write team resources: Create or modify configuration files such as skills, rules, docs, mcp, and hooks in the repository.
• Push changes: Run teamai push to automatically create a branch and open a Merge Request for team review.
• Review and merge: The administrator or designated reviewer approves and merges the MR on the code hosting platform.
• Automatically sync configuration: Each time a member starts an AI session, the SessionStart hook automatically triggers teamai pull Pull the latest configuration to each local AI tool directory.
• Manual sync: You can also run teamai pull at any time to manually update the local configuration.
• Enable knowledge recall: Run teamai recall enable to deploy the sub-agent, letting AI automatically retrieve from the team knowledge base before tasks.
• Capture experience: After a high-friction session ends, run /teamai-share-learnings to automatically de-privatize the pitfall experience and push it to the team repository.
• View team data: Run teamai dashboard to launch the web dashboard, or teamai digest to generate the team weekly report.

Core advantages of teamai-cli

• Git-native workflow: Manage AI configurations with the push→MR→merge process developers are most familiar with, with zero learning cost.
• Unified compatibility across multiple Agents: One team configuration automatically syncs to 9+ tools including Claude Code, Cursor, and Codex, eliminating repetitive configuration.
• Friction-signal knowledge capture: Automatically detect interruptions, retries, and other friction in sessions, turning individual pitfall experience into shared team assets.
• Unified MCP and Hooks management: Declare once in YAML and automatically convert to each tool's native format, greatly reducing maintenance costs.
• Fine-grained subscription control: Use Roles, Tags, and Sources to enable on-demand distribution and filtering by role, by topic, and across teams.
• Intelligent knowledge recall: Automatically retrieve from the team knowledge base before tasks and automatically skip irrelevant tasks, so AI always has team context support.
• Codebase knowledge graph: Build a cross-repository structured dependency graph based on AST parsing, improving the accuracy of AI code understanding and recall.

Project address of teamai-cli

• GitHub repository：https://github.com/Tencent/teamai-cli

Comparison of teamai-cli with similar competing products

Application scenarios of teamai-cli

• Unified configuration management for multi-tool teams: Some people on the team use Claude Code, some use Cursor, and some use Codex. Administrators maintain one configuration in the TeamAI repository, and everyone syncs automatically, ensuring "the same coding standards, the same set of MCP tools, and the same batch of safety hooks."
• Fast AI onboarding for new team members: New members only need teamai init You can automatically get the team's accumulated skills, rules, and best practices, with no need to manually configure each tool one by one, greatly reducing the learning cost of "training the AI".
• Unified security and compliance management: In hooks/hooks.yaml configure the PreToolUse hook, and it automatically takes effect for all team members, avoiding security incidents caused by omissions in individual configurations.
• Automatic accumulation of lessons learned from pitfalls: When a member solves a tricky port conflict problem, the interruptions and retries in the session are captured by friction signals, automatically triggering knowledge accumulation. The next time other members encounter a similar problem, the AI automatically recalls that experience.
• Cross-team skill sharing: Team A publishes public skills to GitHub, and Team B subscribes via teamai source add , enabling cross-team reuse of best practices within the organization and avoiding each team reinventing the wheel.