Content
85%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.
The body is highly actionable with a clear, validated workflow and copy-ready SQL, but it repeatedly defers critical detail (the dual response-key coalesce and submission-dedupe rating-trend SQL) to a references/response-querying.md file that does not exist, undermining progressive disclosure.
Suggestions
Create references/response-querying.md with the copy-ready rating-trend SQL (dual key coalesce + $survey_submission_id dedupe) the body promises, or inline that SQL so the skill is self-sufficient.
Consolidate the repeated "edit, don't re-file" and "memory entry instead of report" guidance into a single stated rule referenced once, to tighten the body.
Verify every referenced path (e.g. references/response-querying.md) resolves to a real bundled file so navigation is never broken.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and assumes Claude's intelligence — no basic-concept padding — but repeats a few rules across contexts ("edit not a new report", "write a memory entry instead") that could be consolidated. | 4 / 5 |
Actionability | Copy-paste-ready SQL, specific MCP tool calls with parameters, exact scratchpad key prefixes, and numeric tiered thresholds give fully executable guidance across the common cases. | 5 / 5 |
Workflow Clarity | A clear orient→explore→memory→decide→close sequence with explicit validation checkpoints (inbox search before authoring, schedule.end_date and addressed: checks) and an edit/author/remember/skip decision tree with feedback loops. | 5 / 5 |
Progressive Disclosure | Structure and one-level-deep reference signaling are good, but the body repeatedly points to references/response-querying.md and no references/ directory exists, so the primary reference target is missing and navigation breaks. | 3 / 5 |
Total | 17 / 20 Passed |