An open-source Collaborative Multi-Agent OS for transparent, human-in-the-loop task coordination via Matrix rooms.
AgentTeams is a runtime platform where multiple AI agents collaborate in shared Matrix rooms under the coordination of a manager. It is for human-supervised or enterprise workflows that need visible, auditable cooperation among agents running on different runtimes, with shared files and centralized traffic management.
This repository also configures its own agents. See what AgentTeams tells them →
Skill Claude CodeCodex
Use before sending @mentions, deciding whether to reply, TASKCOMPLETED, BLOCKED, QUESTION, direct answers, loop prevention, or suppressing low-information acknowledgements.
Skill Claude CodeCodex
Use before direct filesync calls, reading non-task shared files, pushing mid-task progress, or troubleshooting missing shared files. Do not use for normal task acceptance or submission; taskflow acktask and submittask handle lifecycle sync internally.
Skill Claude CodeCodex
Discover and install agent skills from the open ecosystem. Use when you encounter an unfamiliar domain, framework, or workflow that you lack specialized knowledge about, or when your coordinator suggests searching for skills before starting a task.
Skill Claude CodeCodex
Discover and call MCP Server tools via the mcporter CLI. Use when your coordinator notifies you about new MCP tools, or when you need to call external APIs. Includes workflow for generating skill documentation for new MCP servers.
Skill Claude CodeCodex
Use only when you need to look up team topology, worker phase, runtime state, or identity that is NOT available from the current message context. Do not use for standard task flows — the coordinator is the message sender, and the task room is in meta.json.roomid.
Skill Claude CodeCodex
Use before any Worker taskflow call or assigned-task workflow, including reading task state, acknowledging a task, executing a task, tracking progress, handling blockers/questions, submitting structured results, or reporting completion. Always use this skill when the message mentions assigned task, task ID…
Skill Claude CodeCodex
Sync files with centralized storage. Use when your coordinator or another Worker notifies you of file updates (config changes, task files, shared data, collaboration artifacts).
Skill Claude CodeCodex
Use when you are invited to a Project Room or assigned a task within a multi-worker project. Covers project plan reading, task coordination with other Workers, and git author config.
Skill Claude CodeCodex
Use when executing a task (progress logging, plan updates), when resuming a task after session reset, or when managing task history. Covers progress log format, task-history.json, and resume flow.
Skill Claude CodeCodex needs its repo
Execute AI coding CLI tools (Claude Code / Gemini CLI / qodercli / Qwen Code / OpenCode) on behalf of Workers. Use when a Worker sends a coding-request: message, asking the delegating agent (Manager, team Leader, or an authorized Worker) to run coding operations in their workspace.
Skill Claude CodeCodex
Manage the Higress AI Gateway via its Console API (consumers, routes, AI providers, MCP servers). Use when creating consumers, configuring routes, or managing AI gateway settings.
Skill Claude CodeCodex
A procedure for handing coding work to an AI command-line tool, such as Claude Code, Gemini CLI, or qodercli, through a manager agent.
Skill Claude CodeCodex
Use as a helper when the admin is assigning work but has not specified an existing Worker, so you need to search Nacos for a suitable Worker and ask whether to import it; also use when the admin explicitly says to import a Worker from the market or gives a direct nacos package URI.
Skill Claude CodeCodex
Use when determining sender identity in any room, managing trusted contacts, configuring the admin's primary notification channel, handling first-contact from a new channel, or escalating to admin across channels.
Skill Claude CodeCodex
Use when you need to push files to MinIO for Workers to access, pull Worker output from MinIO, or when a Worker reports they've uploaded files. Also use when notifying Workers to file-sync after writing files they need.
Skill Claude CodeCodex needs its repo
Execute git operations on behalf of Workers who don't have git credentials. Use when a Worker sends a git-request: message to clone, push, pull, commit, rebase, or perform any git operation.
Skill Claude CodeCodex
Use when admin requests adding a human user to the system, changing human permissions, removing human access, or managing human accounts.
Skill Claude CodeCodex
Use for standalone Matrix admin requests (register users, create rooms, manage membership, upload files). Also use when you need to send a file to the admin via Matrix media upload. Do NOT use for Worker/project creation — those handle Matrix operations internally.
Skill Claude CodeCodex
Use when admin asks to configure an MCP tool server (e.g., GitHub, weather API), rotate credentials, grant/revoke worker access to MCP tools, or add a custom API integration via YAML.
Skill Claude CodeCodex
Switch the Manager Agent's own LLM model. Use when the human admin requests changing the Manager's model.
Skill Claude CodeCodex
Use when admin asks to start a multi-worker project, when a Worker @mentions you with task completion in a project room, when project plan changes are needed, or when a blocked task needs resolution.
Skill Claude CodeCodex
Expose worker HTTP services via Higress gateway. Use when admin asks to publish a worker's web app or API to make it externally accessible.
Skill Claude CodeCodex
Coordinate access to shared task directories using .processing marker files. Use before accessing a Worker's workspace to prevent conflicts when both Manager and Worker might modify files simultaneously.
Skill Claude CodeCodex
Use when admin requests creating a team, importing a team, managing team composition, adding/removing workers from a team, or delegating tasks to a Team Leader.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: