Tencent Cloud officially announced the open-source release of the team’s AI collaboration tool TeamAI. Built on Git, this tool aims to integrate the team’s skills, work standards, tool configurations, and project knowledge into a Git repository management system. It enables review and version control similar to code management, and is ultimately integrated into each member’s Agent, achieving “one configuration, reused by all”.
Core features and multi-platform compatibility
- Full adaptation of mainstream agents: Currently, TeamAI has been adapted to 16 mainstream agents including WorkBuddy, CodeBuddy, Claude Code, Codex, and Cursor.
- Familiar and efficient Git collaboration: When team members submit skills or modify specifications, the system automatically creates branches and initiates merge requests for team review. After confirmation of integration, the Agent that supports the Hook for session startup will automatically pull the latest updates at the start of a new conversation, eliminating the tedious steps of individual notifications and manual replacements.
- Unified deployment of model configuration and security management: Team administrators can use this tool to uniformly configure model services that meet enterprise security requirements, effectively reducing the risk of members accessing and selecting incorrect models on their own. Additionally, it can prompt sharing of experiences at the end of a session based on correction records and tool failures, and the Agent will organize these experiences into documents and store them in a repository for reference during subsequent task processing.
Integrated AI Platform Management Platform
TeamAI is also now integrated into Tencent Cloud AI Agent Management Platform ClawPro. Members can connect their local Agents to ClawPro, and administrators can uniformly manage configurations such as Skills and MCPs through the control panel, then efficiently distribute them to all connected Agents. This further simplifies the workflow collaboration and standardization processes for teams in the era of large models.