一键导入
agent-teams
Orchestrate teams of Claude Code sessions with shared tasks, inter-agent messaging, and centralized management
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Orchestrate teams of Claude Code sessions with shared tasks, inter-agent messaging, and centralized management
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Complete document processing toolkit - 5 format skills, 3 workflow guides, and 3 templates for PDF, Word, Excel, PowerPoint, and collaborative review.
Multi-AI coordination toolkit router for provider delegation, orchestration patterns, compatibility wrappers, and unified collaboration workflows.
Product management toolkit router with workflow guidance, Lenny-derived PM skills, and reusable PM templates for discovery, strategy, execution, growth, and team leadership.
Interact with Obsidian vaults via the CLI (v1.12.6+). Search, read, create, manage files, tags, properties, tasks, plugins, and more. Use when the user wants to query or manipulate their Obsidian vault programmatically. Requires Obsidian app running with CLI enabled.
Obsidian-centered knowledge management toolkit - 3 format skills, 4 templates, and adjacent productivity skills for knowledge capture, organization, and visualization.
Compatibility wrapper for the legacy PM analytics workflow covering metrics, financial modeling, dashboards, and platform health.
| name | agent-teams |
| description | Orchestrate teams of Claude Code sessions with shared tasks, inter-agent messaging, and centralized management |
| version | 1.0.0 |
| triggers | ["agent team","agent teams","spawn team","create team","teammates","team lead","parallel agents","coordinate agents","multi-agent"] |
Coordinate multiple Claude Code instances working together as a team, with shared tasks, inter-agent messaging, and centralized management.
Agent teams are experimental and disabled by default. Enable them first:
// In settings.json
{
"env": {
"CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS": "1"
}
}
Or set environment variable: CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1
| Aspect | Subagents | Agent Teams |
|---|---|---|
| Context | Own context; results return to caller | Own context; fully independent |
| Communication | Report back to main agent only | Teammates message each other directly |
| Coordination | Main agent manages all work | Shared task list with self-coordination |
| Best for | Focused tasks where only result matters | Complex work requiring discussion |
| Token cost | Lower | Higher (each teammate = separate instance) |
Rule of thumb: Use subagents for quick focused workers. Use agent teams when teammates need to share findings and coordinate.
Tell Claude to create a team in natural language:
Create an agent team to explore this CLI tool design from different angles:
one teammate on UX, one on technical architecture, one playing devil's advocate.
Claude will:
All teammates run inside your main terminal.
Each teammate gets its own pane (requires tmux or iTerm2).
// settings.json
{
"teammateMode": "tmux" // or "in-process" or "auto"
}
Or per-session: claude --teammate-mode in-process
Create a team with 4 teammates to refactor these modules in parallel.
Use Sonnet for each teammate.
Spawn an architect teammate to refactor the authentication module.
Require plan approval before they make any changes.
Press Shift+Tab to cycle into delegate mode. The lead focuses only on coordination (spawning, messaging, task management) without implementing tasks itself.
Assign the security review task to the researcher teammate
Ask the researcher teammate to shut down
Clean up the team
Important: Always use the lead to clean up. Shut down all teammates first.
| Component | Role |
|---|---|
| Team lead | Main session that creates team, spawns teammates, coordinates |
| Teammates | Separate Claude instances working on assigned tasks |
| Task list | Shared work items that teammates claim and complete |
| Mailbox | Messaging system for inter-agent communication |
~/.claude/teams/{team-name}/config.json~/.claude/tasks/{team-name}/Tasks can have dependencies - blocked tasks auto-unblock when dependencies complete.
Teammates don't inherit lead's conversation history. Include details in spawn prompt:
Spawn a security reviewer teammate with the prompt: "Review the authentication
module at src/auth/ for security vulnerabilities. Focus on token handling,
session management, and input validation. The app uses JWT tokens stored in
httpOnly cookies. Report any issues with severity ratings."
Aim for 5-6 tasks per teammate.
Break work so each teammate owns different files. Two teammates editing the same file leads to overwrites.
Check in on progress, redirect approaches that aren't working, synthesize findings as they come in.
If lead starts implementing instead of waiting:
Wait for your teammates to complete their tasks before proceeding
Create an agent team to review PR #142. Spawn three reviewers:
- One focused on security implications
- One checking performance impact
- One validating test coverage
Have them each review and report findings.
Users report the app exits after one message instead of staying connected.
Spawn 5 agent teammates to investigate different hypotheses. Have them talk to
each other to try to disprove each other's theories, like a scientific
debate. Update the findings doc with whatever consensus emerges.
Create an agent team to implement the new notification system:
- One teammate for the backend API
- One for the frontend components
- One for writing tests
Each should own their own files to avoid conflicts.
which tmux returns a pathPre-approve common operations in permission settings before spawning teammates.
Check output with Shift+Up/Down, then:
Tell it to keep going or wait for teammates to finish.
tmux ls
tmux kill-session -t <session-name>
/resume and /rewind don't restore in-process teammatesAgent teams use significantly more tokens than single sessions. Each teammate has its own context window. Use for:
Avoid for routine tasks where single session suffices.