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 highly actionable, well-organized single-file skill with concrete commands, complete response/error documentation, and safety-gated workflows including pre-run checks and verification steps. Its weaknesses are repetition (duplicated curl block, repeated 72-hour and @latest reminders), a boilerplate "When to Use" section, and no error-recovery guidance despite an error table.
Suggestions
Remove the duplicated curl POST block: keep it only in the REST API section and have Quick Start link to it, and drop the boilerplate "When to Use" section, which adds no information.
State the 72-hour claim reminder and the @latest --yes rationale once each (e.g., in Output Checklist and the CLI section) instead of repeating them across the workflow paths.
Add a short error-recovery step to the Agent Workflow (e.g., what to do on a 400/500 API response or a failed CLI run) to turn the documented error table into an actionable feedback loop.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly dense, useful reference material, but the identical curl POST command is duplicated verbatim in Quick Start and REST API, the "@latest --yes" advice and the 72-hour claim reminder each repeat 3-4 times, and the trailing "When to Use" section is empty boilerplate ("Use this skill when tackling tasks related to its primary domain or functionality as described above"). This matches the mostly-efficient-but-could-be-tightened anchor. | 3 / 5 |
Actionability | Fully executable, copy-paste-ready guidance throughout: exact curl commands with payloads, a concrete response JSON shape, an HTTP error table with messages, a complete CLI flags table, a working SDK snippet, and a psql seed command covering the common cases. | 5 / 5 |
Workflow Clarity | Both the API and CLI paths are clearly sequenced with real checkpoints (pre-run .env check, confirm-intent step, verify the connection string was written, Output Checklist) and destructive seed SQL is gated by confirmation. It falls short of the top anchor only because there is no explicit error-recovery loop (the error table lists HTTP codes but no fix/retry guidance) — a minor validation gap, matching the most-checkpoints-present anchor. | 4 / 5 |
Progressive Disclosure | No bundle files exist and the skill is a single well-sectioned ~220-line document with an internal anchor link ("see [Which Method?](#which-method) below") and one external docs link. Structure and navigation are good, but secondary content (full CLI options, SDK, Vite plugin, limits tables) that suits reference files is inlined in the main file, matching the good-structure-with-minor-gaps anchor rather than a well-split overview. | 4 / 5 |
Total | 16 / 20 Passed |