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 tight, well-structured instruction skill: concrete checklists, specific thresholds, a copy-paste output template, and an escalation feedback loop. It falls just short of maximal scores because checks are judgment-based rather than executable, there is no explicit pre-output verification checkpoint, and the opening line redundantly restates the description.
Suggestions
Add an explicit validation checkpoint before emitting the review (e.g., 'Re-verify each flagged coordinate and source URL before listing findings') to close the workflow feedback loop.
Delete the redundant 'This skill goes beyond schema validation to check data quality and plausibility.' line in the body, which restates the frontmatter description.
Make at least one check executable by naming a concrete tool or command — e.g., how to spot-check a coordinate against a geocoding API or the contributions/ schema path — so guidance is not purely judgment-based.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and well-organized as bullet checklists that assume Claude's competence, but it restates 'goes beyond schema validation' from the description (line 16) and carries a few illustrative parentheticals (e.g., the Springfield-in-India example) that could be trimmed — efficient with minor over-explanation. | 4 / 5 |
Actionability | Provides concrete, actionable guidance — specific suspicious patterns, source tiers (Tier 1/Tier 2), an escalation threshold (>100 records), a maintainer handle (@dr5hn), and a copy-paste-ready output template — but the plausibility checks remain judgment-based ('Does the city name make sense') rather than executable steps, leaving minor gaps. | 4 / 5 |
Workflow Clarity | A clear six-step sequence (read context → plausibility → suspicious patterns → verify source → completeness → overall quality) with per-step checklists and an escalation path acting as a feedback loop, but there is no explicit verify-before-output checkpoint re-confirming flagged findings before emitting the review. | 4 / 5 |
Progressive Disclosure | Well-organized with clear section headers and no nested references, and all content is appropriately inline for a self-contained review checklist; at ~95 lines it sits just above the simple-skill threshold and lacks the one-level-deep external references the top anchor expects, so it is well-structured but not maximal. | 4 / 5 |
Total | 16 / 20 Passed |