Skip to main content

Tencent Cloud Open-Sources TeamAI: 16 Agent Skills, 76% Cheaper Code Reviews

Tencent Cloud Open-Sources TeamAI: A Git-Based Hub for AI Agent Skills

On October 10, Tencent Cloud released TeamAI (github.com/Tencent/teamai-cli), an open-source collaboration tool that's already turning heads on GitHub Trending. The project, tested internally at Tencent Cloud, aims to solve a messy problem: how to share and manage AI agent skills across a team without endless copy-pasting.

What TeamAI Actually Does

Think of TeamAI as a Git repository for your team's AI know-how. It unifies skills, work standards, tool configurations, and project knowledge in one place. These assets can be reviewed, updated, and version-controlled just like code, then synced to each member's agent. Configure once, reuse everywhere.

It currently supports 16 agents, including WorkBuddy, CodeBuddy, Claude Code, Codex, and Cursor. When a member submits a new skill, TeamAI automatically creates a branch and opens a merge request for review. After merging, agents with a session-start hook pull the update at the beginning of a new conversation. The shared scope also covers MCP tool configs, rules, and sub-agents. Recently, the team added centralized model configuration distribution.

Image

Learning from Mistakes, Automatically

One standout feature: TeamAI nudges members to share experiences when they hit repeated corrections or failed tool calls. The agent then organizes that info into documents for future retrieval. It's like a team wiki that writes itself.

Image

The Numbers That Matter

In an official test with 13 R&D tasks, a low-cost model that incorporated team knowledge scored 6.1% higher overall. More impressively, model call costs dropped by 76%. That's not just a marginal gain—it's a significant cut in operational expenses.

Beyond the CLI

TeamAI has also been integrated into ClawPro, Tencent Cloud's intelligent agent management platform. So if you're already using ClawPro, you can tap into TeamAI's capabilities without leaving the ecosystem.

Key Points

  • Open-source and free: TeamAI is available on GitHub under Tencent's organization.
  • 16 agent integrations: Works with popular AI coding assistants like Codex, Cursor, and Claude Code.
  • Git-based workflow: Skills are version-controlled, reviewed, and distributed like code.
  • Cost savings: 76% reduction in model call costs in internal tests.
  • Auto-documentation: Prompts members to capture lessons learned and organizes them for reuse.

For teams juggling multiple AI agents, TeamAI could be the glue that keeps everyone on the same page—without the usual chaos.