Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is a well-sequenced, highly actionable workflow with strong validation checkpoints and clear tool/field guidance for an instruction-only skill. It is mostly concise, though a few mechanics are restated across sections, and progressive disclosure is limited by the absence of any bundle/reference files to offload detailed manifests.
Suggestions
Dedupe the 'query carried verbatim preserves variables' point: state it once (step 2 or Important notes) and reference it elsewhere rather than restating it in steps 2, 6, and Important notes.
Tighten step 5's materialisation rationale: keep the safe default and the non-materialisable caveat, but trim the comparative narrative to a line or two.
Consider extracting the full endpoint field manifests (capture list in step 2 and create list in step 6) into a reference file so SKILL.md stays a lean overview with one-level-deep references.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly lean and it assumes Claude's competence (no primers on PostHog/HogQL), but the 'Important notes' section and step 6 restate mechanics already covered in steps 2 and 5 (e.g. copying query verbatim carries variables), and step 5's safe-default rationale is somewhat discursive. Not a 3 because a few passages restate prior content and could be tightened. | 2 / 3 |
Actionability | Every step names the exact tool and the concrete field set to capture and pass (name, query, description, data_freshness_seconds, is_materialized, tags) plus precise naming rules ('Snake_case, URL-safe, starts with a letter, max 128 chars'). Not a 2 because there is no vague or abstract guidance; the instruction-only skill is fully actionable without code. | 3 / 3 |
Workflow Clarity | An explicit 8-step sequence with validation checkpoints: confirm active project is source before reading (step 1), name-collision check before creating (step 3), present summary before copying (step 2), and a verification run (step 7). The 'No undo' create op has a feedback loop via the collision check and 'stop and ask' handling, so it is not capped at 2. | 3 / 3 |
Progressive Disclosure | Well-organized into clear sections with one-level conceptual cross-links to sibling skills and no nested file references, so it is not a 1. But no bundle files exist and the skill is a single ~120-line file with detailed reference-like content (full field manifests, materialisation caveats) inline rather than split into reference files, so it does not reach the well-signaled multi-file structure of a 3. | 2 / 3 |
Total | 10 / 12 Passed |