What problem does it solve? After running a mechanical project scaffolder, the generated CLAUDE.md, README, .docs set, and Claude skill kit still contain placeholder tokens and [GROUND: ...] markers that require actually reading the project's code to fill correctly. This Skill performs that one-time intelligent grounding pass so no placeholder content survives into daily development. ## Core Features & Use Cases - Token and marker resolution: Sweeps for remaining content tokens in CLAUDE.md, README, and .docs, and resolves every [GROUND: ...] marker across the core skills using facts read from the real codebase. - Optional skill enablement: Evaluates prerequisites (CI workflows, Playwright, typecheck configs, phpstan) and moves qualifying optional skills into the active skills directory. - Audit and boot-verify loop: Runs the skill audit until PASS, then boot-verifies the project via setup.ps1, just start/build, and probe checks, feeding real symptoms into troubleshooting docs. - Use Case: After cloning the project-skeleton and running init.ps1 for a new static-site app, invoke the grounding pass so Claude reads the actual routes and justfile, fills every doc placeholder, enables the applicable optional skills, and verifies the app boots before handing off for review. ## Quick Start Tell Claude to run /ground-project after init.ps1 completes so it reads the real code and finishes the scaffolding kit.
