Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-sequenced, highly actionable workflow with specific tool calls and confirmation gates. Its main weaknesses are mild redundancy between the diagnosis, fix-list, and closing-principles sections, and a long single-file layout that makes no use of reference files for progressive disclosure.
Suggestions
Tighten conciseness by collapsing the closing 'What makes a page AI-cite-worthy' section into the Step 4/5 criteria it restates, and trim the motivational opener.
Move the per-CMS tool-name tables (Step 0 discovery and Step 7 push guidance) into a references file (e.g. references/cms-push-guide.md) and link to it, so SKILL.md stays an overview.
Add an explicit post-push verification step (e.g. confirm the draft renders correctly in the CMS / re-fetch the document) to turn the current one-way push into a validate-fix loop.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient and actionable, but the motivational opener ('it's faster to improve a page that already has some signal than to build from scratch') and the closing 'What makes a page AI-cite-worthy' section restate criteria already covered in Step 4/5, so it could be tightened; not padded with basic-concept explanation, so above 1. | 2 / 3 |
Actionability | Names specific tools with parameters (`get_ai_visibility_pages` with `orgBrandId`, `mentionsBrandId`, `sortBy: "citationCount"`; `get_ai_visibility_prompts`; `patch_document_from_markdown`) and a concrete before/after patch template, giving copy-paste-ready executable guidance at level 3. | 3 / 3 |
Workflow Clarity | Steps 0-7 are clearly sequenced with explicit confirmation gates (user picks a page in Step 3, agrees to the diagnosis before rewriting in Step 4, confirms the CMS document ID, and the push is explicitly draft-only with 'Never use publish_documents without explicit instruction'); the draft push is non-destructive so the destructive-cap does not apply. | 3 / 3 |
Progressive Disclosure | Well-organized into numbered sections but everything is inline in a ~270-line SKILL.md with no reference files; content that could be split (the per-CMS tool tables and the closing cite-worthy principles) lives inline, and the under-50-line simple-skill exception does not apply, so it sits at level 2 rather than 3. | 2 / 3 |
Total | 10 / 12 Passed |