Coordinate Claude Code sessions into managed agent teams for parallel work and shared tasks.
npx skills add https://github.com/hlibkoval/claudemd --skill agent-teams-doc
Please help me install this Agent Skill. Skill: agent-teams-doc Source: https://github.com/hlibkoval/claudemd/tree/main/skills/agent-teams-doc Command: npx skills add https://github.com/hlibkoval/claudemd --skill agent-teams-doc
Coordinating multiple Claude Code sessions to work together on a project is complex without built-in orchestration. This Skill provides official documentation for agent teams, enabling parallel exploration, shared task management, secure inter-agent communication, and lifecycle controls.
Spawn a team with three teammates and a lead to tackle a complex debugging task in parallel.
High-intent search queries and answers about installing and using this skill.
To coordinate parallel Claude Code sessions, you can use a managed agent team with a lead to spawn teammates, a shared task list, and a mailbox for secure inter-agent communication. This enables parallel code reviews and cross-layer coordination.
An agent team in Claude Code is used to parallelize work across multiple sessions. It is best suited for tasks that benefit from collaboration, such as architecture planning, debugging with competing hypotheses, and cross-layer coordination.
Yes, running agent teams requires Claude Code v2.1.32 or later. You also need a running lead to spawn teammates, a shared task list, and local storage allocated under ~/.claude/teams and ~/.claude/tasks.
You manage Claude Code teammates using a shared task list and mailbox for inter-agent communication. Best practices include using hooks to enforce quality gates and configuring permissions to manage complex collaborations securely.
Claude Code agent teams support in-process and split-pane display modes. These modes allow you to visually monitor parallel work and shared results across multiple teammates within your development environment.
You should avoid using agent teams for tasks that do not benefit from collaboration or parallel execution. If your task does not require cross-layer coordination or competing hypotheses, the overhead of managing a shared task list and multiple sessions may be unnecessary.
shuff57Coordinate Claude Code agent teams with shared task lists and communication channels.
CaptainCrouton89Coordinate multiple Claude Code sessions to manage agent teams.
hanlinlibhamConfigure OpenClaw multi-agent deployments with four specialized agents.
hidonguyenCoordinates multiple Claude Code agent sessions for parallel research, implementation, review, and debugging.
