How this skill is triggered — by the user, by Claude, or both
Slash command
/code-quality:cleanupThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Scan the codebase for cleanup opportunities, apply safe fixes automatically, and present riskier changes for user
Scan the codebase for cleanup opportunities, apply safe fixes automatically, and present riskier changes for user approval before acting.
Discover source and test files: Glob all source files and test files. Read each one in full before making any changes.
Scan for cleanup opportunities across these categories:
Safe to auto-fix (apply without asking):
TODO/FIXME comments that reference a resolved issue number (e.g., // TODO #42 fix this where #42 is closed)Needs approval (present to user before changing):
Surface only, do not touch (report to user and stop):
TODO/FIXME comments that do not reference a resolved issue — list file, line, and the comment text so the
user is awareRun the formatter if one is configured for the project (e.g., ./gradlew ktfmtFormat,
npx prettier --write ., gofmt -w ., cargo fmt). Skip this step if no formatter is configured.
Apply safe fixes: Make all auto-fix changes directly. Keep each change minimal — do not reformat surrounding code that wasn't already being touched.
Present approval-needed changes: For each item in the "Needs approval" category, show:
Ask the user to confirm which items to apply (e.g., "Apply items 1, 3? All? Skip all?"). Apply only the approved items.
Verify: Run the test suite (e.g., ./gradlew test, npm test, go test ./..., cargo test) to confirm
nothing was broken by the changes. If tests fail, revert the change that caused the failure and report it.
Report: Summarize:
TODO without either fixing the underlying issue or confirming with the user.npx claudepluginhub mattbobambrose/mattbobambrose-claude-skills --plugin code-qualityDetects and auto-fixes AI slop in code (debug leftovers, placeholders, hardcoded credentials) using a three-phase certainty scan, safely applying only deterministic HIGH fixes and verifying with tests.
Detects dead code, fallbacks, deprecated patterns in repos using scanner, git historian, and verifier agents. Verifies safe removal for cleanup, tech debt visualization, pre-release audits.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.