What problem does it solve? Plans often use vague or conflicting domain terminology that drifts away from a project's documented language and recorded architectural decisions, causing design defects and documentation decay. This Skill runs a disciplined one-question-at-a-time interview that challenges a plan against the project's CONTEXT.md glossary and docs/adr/ decision records, updating those files inline as terms and decisions crystallize. ## Core Features & Use Cases - Docs-anchored grilling: Challenges every term in a plan against the existing CONTEXT.md glossary, sharpens fuzzy language, and cross-references claims against the actual codebase. - Inline documentation updates: Edits CONTEXT.md the moment a term is resolved and writes ADRs only when decisions are hard to reverse, surprising, and the result of a real trade-off. - Stdlib-only validators: Ships three Python scripts — a CONTEXT.md linter, an ADR scanner (numbering, format, supersession links), and a glossary-to-code consistency checker that flags dead glossary terms and undefined code-only nouns. - Use Case: Before implementing a new billing feature, run a grilling session to discover that your plan says "account" while CONTEXT.md defines "Customer", resolve the conflict, update the glossary, and record an ADR for the event-sourced write model decision. ## Quick Start Ask the assistant to grill the current feature plan with docs against the project's CONTEXT.md and ADR records.
