Open‑Source AI Collaboration Tool from Tencent Syncs an Entire Team’s Skills with a Single Command

Problem: fragmented AI tool configurations

When multiple developers use different AI coding assistants such as Claude Code, Codex, CodeBuddy, or WorkBuddy, each person stores their own skills, rules, and configuration files locally. Sharing a useful skill or updating a rule requires sending files or chat messages, and newcomers must manually collect each configuration, leading to outdated or inconsistent setups.

Solution: teamai‑cli

teamai‑cli is a Git‑driven command‑line tool that treats AI configuration files as ordinary source code. All team‑wide skills, rules, documentation, and hooks are placed in a shared Git repository, and the same push‑review‑pull workflow used for code is applied to AI settings.

Core workflow

Developers modify a skill or rule locally, run teamai push, which creates a branch and a merge request. After review and merge into the main branch, teammates automatically receive the latest version the next time they start an AI session, because a hook triggers a pull.

Key features

Unified Git repository : All AI tools (Claude Code, Codex, CodeBuddy, WorkBuddy, Cursor, etc.) read their configuration from the same repo.

Shared hooks and MCP service config : Team‑level pre‑commit checks (e.g., secret‑key scanning) and MCP server settings are defined once and distributed to every member.

Pitfall capture : After an AI session ends, the tool scores the session’s friction (interruptions, failed tool calls, retries). High‑friction sessions trigger a prompt to record the experience as a knowledge‑base entry.

Recall before execution : When a new task is started, the AI first searches the team knowledge base for relevant past experiences, documents, or rules, and skips the search if the task is unrelated.

Codebase import : teamai import parses one or more code repositories into structured knowledge, exposing components, interfaces, configurations, and dependencies, so AI can locate relevant source files directly.

Flexible initialization : Teams can create a dedicated team repository or turn an existing project repository into a team repo with teamai init .. The .teamai/ directory stores all configuration and knowledge files.

Subscription to other teams : Public skill repositories from other teams can be subscribed to, allowing reuse of ready‑made configurations.

Getting started

Install the CLI (Node.js and Git are required): npm install -g teamai-cli Create a shared repository on GitHub or an internal Git platform, grant write access, then run in the project directory:

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

The init command handles login, repository linking, member registration, and hook injection. After that, each AI session automatically synchronizes the latest configurations.

For a quick personal trial, run teamai init . in any folder; the tool will create a repository if none exists. A template repository with pre‑populated skills and rules can be used via the “Use this template” button.

Teams that benefit

Teams mixing multiple AI coding assistants and needing a single source of truth for skills and standards.

Teams with frequent onboarding or many contractors, where a one‑click repository clone replaces manual document distribution.

Teams that want to preserve and surface past troubleshooting experiences.

Teams with compliance requirements that need uniform pre‑commit checks such as secret scanning.

Author’s assessment

The problems solved by teamai‑cli are real and will grow as AI coding tools become more widespread. By embedding configuration management into the familiar Git workflow, the tool offers a pragmatic, low‑cost entry point without inventing a new platform. For teams already using AI assistants and struggling with divergent configs or lost knowledge, the author recommends a trial run; the MIT‑licensed project can be installed and removed in minutes with teamai uninstall.

Open‑source repository

https://github.com/Tencent/teamai-cli