Content
78%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 well-structured, domain-rich skill body: lean checklists, a clear 7-step workflow with failure guidance, and exemplary progressive disclosure into two genuine reference files. The main improvement opportunities are tightening the repeated data-gap policy statement and specifying how angles are scored in the workflow.
Suggestions
Specify the scoring model for workflow step 4 (e.g., a simple lead/trail/parity rating per angle or a 1-5 scale with criteria), since "Score each angle" is currently the least actionable step.
Condense the "If required data is unavailable" section to a single statement of the rule (what was needed vs. obtained, gaps stated, no fabrication) to reclaim tokens.
Add one validation checkpoint before data analysis, e.g., confirm the 3-5 selected competitors show real SERP overlap before pulling exports.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is largely lean — bullet checklists of concrete investigative questions, a compressed 7-step workflow, and failure patterns that convey real domain insight without explaining concepts Claude already knows. But a few spots could be trimmed: the "If required data is unavailable" section states its single rule three ways ("the sanctioned output is the deliverable with the gap stated... Fabricating... is never sanctioned. A stated gap is a complete answer."), and the tool paragraph ends with the somewhat padded "Each platform gives a different cut of the same competitive landscape; using them together produces sharper signal than any one alone." That is the score-4 anchor (efficient with minor over-explanation that could be trimmed), not 5. | 4 / 5 |
Actionability | Guidance is largely concrete and executable for an instruction-only skill: exact competitor-derivation procedure ("listing the top 10 ranking domains for the user's top 10 priority keywords. Pick the 3 to 5 that appear most often"), specific metrics per angle (LCP/INP/CLS, "Top 50 referring pages by traffic value", domain rating), named tools, a linked report template, and a defined output structure. The gap is step 4's "Score each angle" — no scale or scoring method is stated in the body, and prioritization criteria are only a parenthetical — which fits the score-4 anchor (mostly executable with minor gaps) rather than 5. | 4 / 5 |
Workflow Clarity | The 7-step workflow is clearly numbered and sequenced, each of the five angles has an investigation checklist, and the "If required data is unavailable" section gives an explicit error path plus a failure-patterns section covering common mistakes. However there are no validation checkpoints on intermediate outputs (e.g., verifying exports before scoring, confirming competitor selection before pulling data), so it sits at the score-4 anchor (clear sequence, most checkpoints present, minor validation gaps) rather than 5. It is well above 3 since the sequence is complete and failure handling is explicit. | 4 / 5 |
Progressive Disclosure | SKILL.md is a true overview: the 5-angle framework and workflow live inline at the right altitude, while the fillable template and the detailed content-gap methodology are split into two real, clearly signaled one-level-deep references (linked in the workflow step, in the Reference files section with descriptions, and verified to exist with substantial relevant content). No nested-reference chains and no content that should have been extracted. This matches the score-5 anchor exactly. | 5 / 5 |
Total | 17 / 20 Passed |