Content
67%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 exceptionally actionable and grounded in project-specific constraints Claude could not infer, with strong validation checkpoints, but it is a monolithic 1,000+ line reference with no progressive disclosure: five major reference topics that belong in separate bundle files are inlined, and one large worked example duplicates earlier sections. Splitting it into an overview plus reference files would fix both structure and token cost.
Suggestions
Move the SubBlock Types Reference, the Canvas Sentences contract, and the BlockMeta/templates/skills patterns into files under references/ (e.g. references/subblock-types.md, references/canvas-sentences.md, references/block-meta.md) and keep SKILL.md as a concise overview linking to them — currently all five major reference topics are inlined in a single ~1,050-line body with no bundle files at all.
Delete or heavily trim the ~90-line "Complete Example": it duplicates the Block Configuration Structure template and the subBlock, conditions, tools, inputs, and outputs patterns already shown in their own sections.
Consolidate the duplicate-subblock-id and canonicalParamId constraint rules, which are currently stated in "SubBlock Types Reference", "File Input Handling" (Constraints), and "canonicalParamId Pattern", into one authoritative section referenced from the others.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Most content is genuinely non-obvious project-specific knowledge (serializer internals, generator commands, check-script behaviors) that earns its tokens, but the ~90-line "Complete Example" largely duplicates the Block Configuration Structure and subBlock patterns, and the duplicate-id/canonicalParamId constraints are restated across multiple sections. Mostly efficient but could be tightened, matching anchor 3. | 3 / 5 |
Actionability | The body provides copy-paste-ready TypeScript config templates, exact commands (e.g. `bun run generate:deployment-config`, `bun run apps/sim/scripts/check-canvas-sentences.ts --block={service}`), exact file paths, and explicit good/bad code contrasts covering the common cases. Fully executable, matching anchor 5. | 5 / 5 |
Workflow Clarity | A brief three-step task sequence opens the skill, and the "Final Validation (Required)" section plus the ~30-item "Checklist Before Finishing" provide explicit validation checkpoints with runnable check commands. However, the actual execution order (build → regenerate artifacts → run checks) must be assembled from scattered reference sections, leaving minor sequence gaps consistent with anchor 4. | 4 / 5 |
Progressive Disclosure | The skill is a single monolithic ~1,050-line body with no references/, scripts/, or assets/ bundle files. Content that clearly belongs in separate files — the SubBlock Types Reference, the Canvas Sentences contract, the BlockMeta/templates/skills patterns, the OAuth deployment gating procedure — is fully inlined. Section headers do exist (so it is navigable, not anchor 1), but the inlined volume far exceeds the anchor-2 example's "300 lines that should be in a separate file". | 2 / 5 |
Total | 14 / 20 Passed |