What problem does it solve?
The launch Skill removes uncertainty and setup friction by turning an ambiguous or structured project request into a concrete execution plan with environment, tooling, testing, and launch orchestration.
Core Features & Use Cases
- Task analysis and environment inventory: Classifies the work type (spec/build/fix/refactor/research), estimates complexity, and checks for existing MCPs, hooks, skills, feature tracking, evals, and init scripts.
- Approach and tooling recommendations: Proposes single-session vs sub-agents vs agent teams, including concrete suggestions like Stitch MCP for UI mockups and Playwright MCP for UI testing.
- Structured project setup and guarded launch: Generates
.claude/features.json, .claude/progress.md, .claude/evals/scenarios/, .claude/settings.json hooks, and a project CLAUDE.md, then runs a pre-flight checklist before launching Phase 4 execution.
Quick Start
Use the launch skill by telling Claude Code to run /launch with your project requirements or spec so it can analyze the task, propose an approach, set up .claude/ scaffolding, and then launch the work after your approval.