Lean, action-oriented body with minimal padding; assumes Claude's competence and relies on hard tables and copy-paste scripts rather than restating known concepts, though a couple of philosophy points are repeated across sections.

Provides fully executable Phase D and E scripts, a concrete wrong-to-correct fix table, and exact shell commands (e.g. the rg invocation), all copy-paste ready.

Sequenced phases with time estimates, an explicit validation checklist, and clear feedback loops ('stop, fix it, re-run... do not proceed until clean') for batch/destructive doc operations.

References are signaled in a Reference Files section but API_REFERENCE.md, DOCS_STRUCTURE.md, and scripts/test_doc_code_blocks.py do not exist in the bundle, and large executable scripts remain inline rather than split out.

Lists multiple concrete actions (find, fix, remove) with specific failure types (wrong code, broken links, incorrect counts) plus an explicit five-phase breakdown, matching the 'lists multiple specific concrete actions' anchor.

Explicitly answers both what (find and fix/remove wrong, outdated, or redundant info in ToolUniverse docs) and when (explicit 'Use when...' trigger guidance).

The 'Use when reviewing docs, before releases, after API changes, or when asked to audit, fix, or simplify documentation' clause covers natural terms a user would actually say.

Tightly scoped to ToolUniverse documentation QA with project-specific phase letters and terminology, giving it a clear niche unlikely to trigger for unrelated skills.