Content
75%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A disciplined, well-structured checklist skill: dense actionable rules, priority ordering, and clean sectioning with no filler. It falls short of top marks on minor redundancy between the workflow and review-guidance sections, unspecified verification methods, and the absence of any feedback loop for failed checks.
Suggestions
Merge or de-duplicate the 'review guidance' section against the Workflow steps — four of its five bullets already appear verbatim in the workflow above.
Make verification concrete: name how to check social cards (e.g., the Open Graph debugger / Twitter card validator on the real URL) and how to detect duplicate title/description/canonical tags (e.g., grep for repeated tags in the rendered head).
Add a feedback loop after the verification step: what to re-check and in what order when cards fail to render or the tags disagree, so fixes converge instead of ending the workflow at a failed check.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean rule-bullets with no concept-explanation padding, but the 'review guidance' section restates workflow steps 3–5 (agreement check, real-URL verification, minimal diffs), creating trimming opportunities that keep it below the top anchor. | 4 / 5 |
Actionability | Most rules are concrete, checkable directives ('Open Graph and Twitter images must use absolute URLs', 'set twitter:card... summary_large_image by default'), but verification steps name no tool or method and 'escape and sanitize' offers no technique, leaving minor gaps. | 4 / 5 |
Workflow Clarity | The 5-step workflow is clearly sequenced by priority with explicit checkpoints (steps 3–4), but there is no feedback loop describing what to do when verification fails or how to re-check after fixes. | 4 / 5 |
Progressive Disclosure | Sections are well organized (workflow, priority table, eight categorized rule blocks) with everything inline and no buried references, but at ~100 lines with no bundle files, the rule catalog could be split into a reference file to keep SKILL.md a leaner overview. | 4 / 5 |
Total | 16 / 20 Passed |