Tencent's open-source TeamAI CLI syncs a team's skills, rules, agents and MCP config from one Git repo, so each person stops keeping their own set
When every teammate keeps their own Claude Code / Codex / Cursor skills, rules and MCP servers, configs drift fast. Tencent's open-source TeamAI CLI (npm package teamai-cli, MIT) puts those AI assets in a shared Git repo: after one teamai init, later sessions automatically pull the latest skills, rules, agents, hooks, MCP and workflows. You can also install and share through the /teamai skill inside an agent. Chinese-community account TryAI Lab (@TryAiLabHQ) ran init → pull → push locally on 10 Nov 2026 with v0.29.0: after pull, .claude/ held 1 skill, 7 rules, 76 docs, 3 agents and 21 workflows, and a hand-written my-review skill pushed back into the team repo. Two real pitfalls: the first push right after init fails until you commit the scaffold files TeamAI generated; pull also printed twenty-plus workflow schema warnings where the team's own main-branch workflows failed validation against CLI 0.29.0. We installed [email protected] from npm (teamai --version matched), confirmed doctor asks for init when uninitialized, and saw init against teamai-hub/teamai-template stop with “GitHub authentication unavailable” in a non-interactive shell without GITHUB_TOKEN. We also cloned the public template and counted 11 SKILL.md files, 11 rules under rules/common, and 3 review agents. We did not finish an authenticated pull/push, and did not open a Claude Code session to verify the auto-sync hook.
Visit site TeamAI CLI (Tencent) ↗- Pricing
- Free
- Region
- Global
- Where it's from
- Chinese community
- Added
- Nov 10, 2026
- Tested on
- Nov 10, 2026
- Version we used
- [email protected] (npm); template teamai-hub/teamai-template; source Tencent/teamai-cli @ e2791c3
- Categories
- CodingProductivityAutomation & Agents
- Original author
- TryAI Lab @TryAiLabHQ ↗
- Our test
- Partly tested
- Cost / time
- The CLI and templates are MIT/free. You need Node/npm, Git, and a team Git host account. Model spend still comes from whatever Claude / Codex (etc.) subscription you already pay; TeamAI does not add a fee.
The task
Share one set of skills, rules, agents and MCP across the team—instead of everyone pasting their own—and keep each session on the latest repo version.
Steps
- You need Node.js (for npm), Git, and an agent you already use (Claude Code, Codex, Cursor, …). The team needs a shared Git repo with branch push access; if you have none yet, fork https://github.com/teamai-hub/teamai-template.
- Install the CLI: npm install -g teamai-cli. Confirm with teamai --version (npm currently served us 0.29.0). Or hand your agent: Install the teamai skill: https://github.com/Tencent/teamai-cli/tree/main/skills/teamai , load it, then set up TeamAI for my team from scratch. The Chinese README also lists a skillhub.cn install line when GitHub is hard to reach.
- Admins run teamai init https://github.com/your-org/your-repo on the shared repo; members run the same from their project (or --scope user for a home install). Non-interactive shells need GITHUB_TOKEN/GH_TOKEN with repo scope, or an already-logged-in gh, or init stops at authentication.
- After init, teamai pull fetches team assets; teamai list skills / rules / agents / mcp shows what landed (often under .claude/ in the workspace). Later agent sessions sync via hooks. Share a local skill back with teamai push (or ask /teamai to share it).
- Use teamai doctor and teamai status when something looks off. The first push right after init fails until you commit the scaffold files TeamAI wrote. If workflow validation warnings flood the pull, keep workflows off for now and sync skills/rules first.
Prompt
Install the teamai skill: https://github.com/Tencent/teamai-cli/tree/main/skills/teamai , load the teamai skill, then set up TeamAI for my team from scratch.
What the author says
@TryAiLabHQ (10 Nov 2026): not another coding agent—an orchestrator that keeps team skills, rules and MCP in Git and syncs them when you open a session. Their run: npm installed v0.29.0 in about 11 seconds for $0; init cloned the repo and wrote a Claude Code SessionStart hook; pull landed 1 skill / 7 rules / 76 docs / 3 agents / 21 workflows; push accepted their my-review skill into the team repo. Two pitfalls: the first post-init push always fails until scaffold files are committed; pull printed workflow validation errors against CLI 0.29.0 on the team's own main. Tencent's claim of +6.1% quality / −76% cost on 13 tasks with a cheaper model is marketing without a published method. Verdict from the post: skills/rules sync is usable; leave workflows alone for now. The official README also lists support for Claude Code, Codex, Cursor, CodeBuddy and more, plus a /teamai conversational setup flow.
Our test
On 10 Nov 2026 on Linux we ran npm install -g [email protected]; teamai --version printed 0.29.0; teamai doctor correctly reported an uninitialized workspace. TEAMAI_MODE=unattended teamai init against https://github.com/teamai-hub/teamai-template stopped at GitHub authentication without GITHUB_TOKEN or an interactive terminal. We also git-cloned that template and Tencent/teamai-cli (e2791c3) for a read-only layout check; we did not npm ci / vitest (source package.json still says 0.22.0 while npm publishes 0.29.0). No live Claude Code/Codex session; no authenticated pull/push.
What we found when we tried it:The global teamai binary works; doctor points at init. The public template has 11 SKILL.md files (ecc + mattpocock and others), 11 rules under rules/common, and three agents (code-reviewer, security-reviewer, database-reviewer); that template has no workflows/docs trees (counts differ from @TryAiLabHQ’s team repo because the repos differ). LICENSE states MIT.
How to get in
GitHub and npm are globally available; READMEs cover Chinese, Japanese, Korean and Thai. When GitHub is hard to reach, the Chinese README lists a skillhub.cn install line. You need a team Git host that allows branch pushes (GitHub, GitLab, GitCode, CNB, TGit, …).
Who it suits
Small teams of two or more already on Claude Code / Codex / Cursor who are tired of shipping skill zips in chat. Skip if you work alone or do not want a shared Git repo.
Pitfalls
- Without GITHUB_TOKEN or a logged-in gh in a non-interactive shell, teamai init fails authentication—we hit that.
- The first push after init often fails until you commit the generated scaffold files.
- 0.29.0 can warn that in-repo workflows have unknown fields; @TryAiLabHQ suggests sticking to skills/rules first.
- Tencent's +6.1% quality / −76% cost claim has no published evaluation detail—treat it as marketing.
- We did not complete an authenticated pull/push, and did not verify the Claude Code auto-sync hook in a live session.
More in this category
- A demo video per docs page that flags itself when the UI changesVideoScreenplayClaude CodeTool free; uses your agent's quota◐ Partly testedOverseas
- AI backend code that guards against races, leaks and bad migrationsCodingbackendskillsClaude CodePostgreSQLFree; about 2× tokens per task◐ Partly testedOverseas
- An afternoon of AI work, a 3D action game you play in the browserCodingClaude CodeTripoJimengSkill free; Claude quota plus asset tools◐ Partly testedChina
- Keep a long-running AI-built project from turning into a messCodingClaudeCodexoil-delegateNeeds Claude and Codex plans◐ Partly testedChina
- Use a web app once and get an AI write-up of how it worksResearchREAClaude CodeChromeFree tools; needs Claude or Codex◐ Partly testedOverseas