Content
100%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.
An exemplary skill body: lean, fully actionable, and sequenced with explicit validation and a completion checklist. It teaches by pointing to a canonical in-repo example rather than inlining or over-explaining, keeping every token load-bearing.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every step is project-specific instruction ("keep them free of the canvas and the DOM so Vitest can run them", "Never use local storage directly") with zero padding and no explanation of concepts Claude already knows. Not 4 because nothing could be trimmed without losing actionable information. | 5 / 5 |
Actionability | Gives concrete file paths ("src/games/<game-id>/" with game.ts, game.test.ts, README.md), exact commands ("npm run format, then npm run check"), specific API calls ("recordScore(ringToss.id, score) from src/carnival"), and a copy-paste TypeScript export template. The comment placeholders inside start() are justified flexibility since each game's implementation differs, not a gap. | 5 / 5 |
Workflow Clarity | Ten numbered steps in clear dependency order, a fix loop ("fix anything it reports"), and a 'Done when' checklist with explicit validation (game playable at #/play/<game-id>, best score shows, "npm run check passes, including the registry test"). Not 4 because the validation checkpoints are explicit and close the loop with success criteria. | 5 / 5 |
Progressive Disclosure | Well-organized sections (intro, Steps, Done when) with no inlined content that belongs in separate files, and it delegates the pattern to a single clearly-signaled source ("Read src/games/water-cannon/game.ts first: it is the pattern to copy") instead of duplicating it. No bundle files exist and none are needed at this size. | 5 / 5 |
Total | 20 / 20 Passed |