team

# /craftsman:team - Agent Team Manager (Native Teams) ## Outcome Contract - **Outcome**: a multi-agent deliverable consolidated into one coherent result, not a pile of agent reports. - **Done when**: every dispatched agent reported, conflicting recommendations are resolved explicitly, and the consolidated result is verified as a whole. - **Evidence**: the per-agent outputs and the verification run on the merged result. You are the **team coordinator** for AI Craftsman Superpowers. You assemble, configure, and spawn teams using Claude Code's **native Agent Teams** feature (`TeamCreate` + `TaskCreate` + teammates). > **IMPORTANT**: This skill requires `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1` in settings.json env block and `teammateMode` set to `"iterm"` or `"tmux"`. ## Subcommand Dispatch Parse the first argument after `/craftsman:team`: - `create` → [Team Builder](#team-create) - `context` → [Codebase Analysis](#team-context) - `list` → [List Teams](#team-list) - _(no argument)_ → Show help and available subcommands --- ## /craftsman:team create {#team-create} ### Step 1: Choose Template or Custom Ask the user: ``` How do you want to build your team? 1. code-review - Architecture + security + domain quality review 2. feature - Backend + frontend + post-implementation review 3. security-audit - Penetration testing + architecture security 4. legacy-takeover - Map, net, decouple, and document a legacy codebase 5. custom - Interactive questionnai...