Lean and token-efficient: concrete thresholds, a scoring table, a decision tree, and failure patterns with no concept-explanation padding of facts Claude already knows, matching the score-3 anchor.

Instruction-only but highly actionable — specific metric thresholds with sources, a decision matrix, explicit spreadsheet columns, and a sequenced workflow; absence of code is not penalized because the guidance is concrete.

Seven-step sequenced workflow with an explicit validation checkpoint ('Re-crawl after each batch to confirm no broken redirect chains') and a post-audit measurement feedback loop (30/60/90 days), plus a Failure patterns guardrail section; validation is present so the destructive/batch cap-at-2 does not apply.

Clear overview kept inline with two well-signaled, one-level-deep references (audit-template.md, cannibalization-resolution.md), both verified to exist; detail is appropriately split, matching the score-3 anchor.

Lists multiple concrete actions — 'keep, update, merge, redirect, or delete' — matching the score-3 anchor for enumerating specific concrete actions rather than naming only a domain.

Explicitly answers both what ('decide what to keep, update, merge, redirect, or delete') and when ('Use this skill whenever the user wants to audit existing content...'), with an explicit 'Use when...' clause.

Broad, natural trigger coverage an SEO practitioner would actually say ('content audit', 'content decay', 'cannibalization', 'prune content', 'redirect old pages', 'evergreen refresh'), matching the score-3 anchor.

Occupies a clear niche (keep/update/merge/redirect/delete framework) with distinct triggers unlikely to fire for adjacent skills; not generic enough to conflict broadly.