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.
A strong, actionable pattern reference with executable TypeScript examples and crisp always/never rules, undermined primarily by a broken progressive-disclosure layer: all referenced files are missing from the bundle, and the inlined detail that should live in them makes the skill heavier than its structure promises. Minor code-syntax errors in the cron section slightly reduce copy-paste reliability.
Suggestions
Ship the referenced files (references/step-execution.md, error-handling.md, observability.md, checkpointing.md) or remove the dead links; fix the "../references/expressions.md" path to stay inside the skill directory.
Move the deep-dive sections (cancellation patterns, logging setup, checkpointing config) into the reference files so SKILL.md is a lean overview, matching the Next Steps promises.
Fix the cron trigger snippets' syntax errors (semicolons inside object literals, mismatched brackets) so every code block is copy-paste executable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and example-driven with almost no padding of concepts Claude already knows; Inngest-specific semantics (memoization, step counters, attempt semantics) genuinely add information. Minor trimmable material exists: the "Core Concepts" bullet list partially duplicates "When to Use Steps", and "Common Mistakes to Avoid" recaps points already made with code. Not 5 because of those few redundant passages; not 3 because there is no noticeable over-explanation. | 4 / 5 |
Actionability | Mostly copy-paste-ready TypeScript covering creation, triggers, idempotency, cancellation, NonRetriableError/RetryAfterError, logging, and checkpointing, with good/bad contrasts. Not a 5 because of real minor gaps: the cron trigger snippets contain syntax errors ("{ cron: \"0 */6 * * *\"; }" uses a semicolon inside an object literal and the combined example mixes "[" and "{"), and a few snippets are fragments ("/* ... */" bodies) or reference undefined helpers without noting them. | 4 / 5 |
Workflow Clarity | This is a pattern-reference skill, and it sequences guidance logically (limits → when to use steps → creation → triggers → idempotency → cancellation → errors → logging → performance) with explicit decision rules ("Always wrap in step.run()" / "Never wrap") and error-recovery patterns (try/catch fallback). Not a 5 because there is no verification step for the resulting functions (e.g., testing with `inngest dev` or validating trigger configs) and no guidance on ordering concerns when mixing steps and conditionals. | 4 / 5 |
Progressive Disclosure | The body points to "references/step-execution.md", "references/error-handling.md", "references/observability.md", "references/checkpointing.md", and "../references/expressions.md", but no references/ directory exists anywhere in the bundle — every file reference is dead, and the expressions.md path even escapes the skill directory. Per the guideline to score against the actual bundle structure, the claimed progressive disclosure is broken, and detailed material (cancellation patterns, logging setup, checkpointing) that the Next Steps section says lives in reference files is instead inlined. Not a 1 because the SKILL.md itself is well sectioned and the Next Steps list clearly signals where detail should live; not a 3 because signaling to nonexistent files fails the core purpose of progressive disclosure. | 2 / 5 |
Total | 14 / 20 Passed |