build_playbook

Structured interrogation that writes every decision to disk with unique IDs, structured records, and traceability. Produces a decision graph that downstream commands (write-a-prd, prd-to-gsd, prd-to-ralph) compile — not re-interview. Use when user wants to stress-test a plan, get grilled on their design, or mentions "grill me".

COMPILE a production-grade PRD from decision artifacts on disk. Reads grill logs, decision indexes, and context files — does NOT re-interview. Every claim traces to a decision ID. Untraced claims are flagged as assumptions. Use when the user wants a PRD that will feed downstream AI builders such as /playbook:prd-to-gsd and /playbook:prd-to-ralph.

Senior PostgreSQL database engineer. Reads .planning/data-requirements.md (from /data-grill) and designs a future-proof database: Drizzle ORM schemas, migration SQL, anti-pattern validation, visual diagrams, and documented decisions. Use AFTER running /data-grill. Use when designing a new database, extending an existing schema, or reviewing database quality. Never use without data requirements — run /data-grill first.

Extract the COMPLETE capability surface of any module or app — including features that came in through libraries you never explicitly asked for. Uses 3 layers: code extraction, library enrichment, and runtime crawl. Use BEFORE writing any test specs, QA documents, or feature docs. Use when you need to know everything a module can do. Use after building any feature.

Senior cloud infrastructure engineer. Reads .planning/infra-requirements.md (from /infra-grill) and designs the complete hosting setup: platform selection, deployment pipeline, monitoring, file storage, CDN, email, and environment management. Generates config files. Explains every decision in plain English. Use AFTER running /infra-grill. Use when setting up hosting for a new project, changing platforms, or scaling infrastructure.