Tencent's teamai-cli syncs AI coding rules across a team

teamai-cli has 2,113 stars and 142 forks. It syncs a team's skills, rules, MCP servers and knowledge from one shared git repo into Claude Code, Codex, CodeBuddy, WorkBuddy, OpenCode and Cursor, at either project or user scope.

Most teams handle this the opposite way. Each developer keeps a personal skill directory and installs whatever looks useful that week, with no review step before it reaches their editor. teamai-cli treats the shared repo as the thing you distribute, with write access granted to teammates instead of to a random GitHub star count.

That's a real fix for a real problem. Skill files pulled from wherever tend to be inconsistent quality, and a repo with review on it catches the bad ones before they reach six different tools at once. One person fixes a broken MCP server config, everyone's setup updates.

What the project page doesn't say is how conflicts get resolved when two team members edit the same skill, or what happens when Claude Code's skill format diverges from Cursor's. Six target tools means six surfaces to keep compatible, and that maintenance cost doesn't show up in a star count.

1,215 stars in one day is 57% of the repo's total. That says a lot of people hit the same problem at once and not much about whether the tool holds up under daily use. I'd want to see it running on a real team's repo for a month before calling the pattern solved.