Content
92%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is lean, highly actionable, and clearly sequenced with validation feedback loops. The only weak spot is progressive disclosure: a longer self-contained skill with dense API/SQL detail that could be factored into a reference file.
Suggestions
Move the autocapture SQL query (Step 3) and possibly the 'Reading the signals' table into a references/ file (e.g. references/signals.md), keeping a one-line pointer in SKILL.md so the overview stays lean and earns a one-level-deep progressive-disclosure structure.
If the querying-posthog-data/models-heatmaps and exploring-autocapture-events pointers are the intended deep references, make them explicit file links (e.g. [references/autocapture-query.sql]) rather than relying on external skill names, so the bundle structure is self-contained and navigable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with domain-specific facts Claude would not know (16x coordinate scaling, unique_visitors vs total_count, fold semantics, async saved-heatmap rendering) and avoids explaining basics like what a heatmap or SQL is; length is substantive instruction rather than padding. | 3 / 3 |
Actionability | Provides an executable autocapture SQL query and concrete tool-call sequences (heatmaps-list with type/aggregation/limit/offset/has_more, heatmaps-saved-create then poll heatmaps-saved-get on short_id/status), matching the copy-paste-ready anchor. | 3 / 3 |
Workflow Clarity | A clear six-step sequence (Step 1 through Step 6) with explicit validation checkpoints — confirm URL if ambiguous, check has_more, check opt-in/URL when results are empty, poll until status is completed — matching the explicit-validation anchor. | 3 / 3 |
Progressive Disclosure | No bundle files exist and the content is self-contained with well-organized sections, but at ~144 lines it exceeds the simple-skill threshold and keeps dense detail (the SQL query, the signals table) inline rather than splitting it into one-level-deep reference files, matching the some-structure-but-inline anchor. | 2 / 3 |
Total | 11 / 12 Passed |