What problem does it solve?
Coordinating multiple independent Claude Code instances manually is slow, error-prone, and hard to track; this Skill enables parallel execution of teammates so engineering work like implementing briefs, reviews, investigations, and refactors completes faster with clear ownership and session tracking.
Core Features & Use Cases
- Spawn Teammates: Create isolated Claude Code teammate sessions that run independent workflows (hunt, review, investigate, refactor) with their own context and tools.
- Ownership & Safety: Enforce file ownership boundaries, detect file conflicts, require experimental flag CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS, and provide clean shutdown and error handling.
- Specialized Reviews & Investigations: Run multi-angle reviews (security, performance, standards) or hypothesis-driven investigations across teammates and synthesize unified reports.
- Session Tracking & Messaging: Maintain ephemeral team state in ai/session/CURRENT_SESSION.md, support messaging/broadcast to teammates, and summarize results on shutdown.
Quick Start
/team hunt FR-022 FR-023 FR-024 to run three briefs in parallel with isolated teammate sessions and file ownership boundaries.