Scrub performs a repository-wide consistency and drift audit covering markdown linting, links, placeholders, line endings, privacy, configuration truthfulness, documentation indexes, TODOs, and template drift. It reports findings without changing files. Use it for repository health checks, restructures, or after moves and renames.
Install
npx skillscat add dracocephalum/dc-agentics/scrub Install via the SkillsCat registry.
Scrub performs a repository-wide consistency and drift audit covering markdown linting, links, placeholders, line endings, privacy, configuration truthfulness, documentation indexes, TODOs, and template drift. It reports findings without changing files. Use it for repository health checks, restructures, or after moves and renames.
A consistency and drift pass. Not a bug hunt - that is the built-in/code-review. Not a rules review - that is /review. Building, testing
and dependency licences are deliberately outside it, so a scrub stays cheap
enough to run without deciding to.
Find the standard
docs/rules/scrub.md- an initialized repositoryrepo/scrub.md- the dc-agentics toolkit, which adds its own checks on top
ofrepo/payload/docs/rules/scrub.md
Read the first that exists and follow it as written. Never scrub from memory:
the checks exist because each one caught something that review did not.
Scope
The whole repository, always. Most checks are repository-wide by nature - theAGENTS.md index, .dc-agentics.yaml truthfulness, TODO.md, baseline drift,
duplication - and a narrowed pass reported as clean is worse than no pass. If$ARGUMENTS names paths, run everything and lead the report with those.
Report
A findings table - number, check, pass / fail / skipped, and detail - with
what would close each failure. Change nothing. Fixes happen only when asked,
as a separate reviewable change, and a baseline drift is a decision for the user
rather than a file to rewrite.