docguard-fix

To fix broken architecture documentation, diagnose the issues, research the actual repository files, and iteratively rewrite the canonical docs with accurate file-path references and project-specific content until validation checks pass.

The best way to generate security docs that match real dependencies is to inspect the repository first, then write content that reflects actual files and modules, applying validation-aware iteration to ensure documentation guards pass.

Guard-aware documentation repair works by using diagnostic and validation feedback to iteratively correct stale test specs, researching the actual codebase to rewrite content until checks pass or the repair limit is reached.

You can repair data model and environment setup docs, but it specifically applies to repositories that use guard-based documentation standards and require structured metadata, accurate file-path references, and section rules.

Generated documentation contains placeholders when the actual codebase is not researched first; this skill prevents that by diagnosing issues, inspecting real repository files, and rewriting canonical docs with project-specific content.

A research-driven codebase approach can repair canonical documents including architecture, security, testing, data model, environment, and requirements documentation by inspecting the repository and matching real files and dependencies.