harness-step2-fill-docs

To generate architecture and conventions documentation automatically, this Skill deep-reads your codebase to populate docs/ files like ARCHITECTURE.md and CONVENTIONS.md with inferred module responsibilities, dependency rules, and naming patterns backed by code examples.

The best way to track tech debt from code analysis is using a structured tracker; this Skill populates exec-plans/tech-debt-tracker.md by extracting discovered issues like duplicate code and TODOs directly from your codebase structure.

To document technology decisions and library rationale, this Skill populates TECH_DECISIONS.md with inferred reasoning for major frameworks and libraries, explicitly labeling uncertain items as pending review for human confirmation.

Before automating project documentation extraction, you need an AGENTS.md file and an existing docs/ skeleton in place; this Skill then reads the codebase to fill those skeleton files with substantive architectural and convention content.

The documentation generation process handles uncertain technical details by labeling inferred items as pending review and surfacing them for human confirmation, ensuring all reference-backed documentation remains accurate and verifiable.

Yes, you can establish a Definition of Done and code review checklist from your codebase; this Skill populates QUALITY.md with these standards by analyzing your project's existing code structure and engineering patterns.