Content
70%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A highly actionable, well-structured reference with excellent executable examples, but it is token-heavy and keeps a full API reference inline rather than progressively disclosing it, and workflows omit explicit validation checkpoints.
Suggestions
Move the verbose API Reference tables into a separate references/ file (e.g. REFERENCE.md) and link to it, keeping SKILL.md as a lean overview to improve progressive disclosure and conciseness.
Add explicit validation/retry checkpoints for the uploadImage, generate, and edit workflows (e.g., catch StitchError, inspect e.recoverable, retry on RATE_LIMITED) to lift workflow clarity.
Trim repeated `import { stitch } from "@google/stitch-sdk"` lines and consolidate the duplicated error-code lists to reduce token usage.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient with executable code, but the large inline API-reference tables and repeated import statements add bulk that could be tightened or offloaded; it assumes Claude's competence but is not consistently lean. | 2 / 3 |
Actionability | Provides fully executable, copy-paste-ready TypeScript examples for every major operation plus concrete method signatures, options tables, and error codes, matching the 'fully executable code' anchor. | 3 / 3 |
Workflow Clarity | Sections are well-labeled and the quick-start sequence is clear, but multi-step generation/iteration/upload workflows lack explicit validation or error-recovery checkpoints, which caps clarity at 2 per the destructive/batch-operation guideline. | 2 / 3 |
Progressive Disclosure | Content is well-organized into sections, but it is a monolithic single-file skill with no bundle references and a large inline API reference that could live in a separate file; no references directory exists to split content one level deep. | 2 / 3 |
Total | 9 / 12 Passed |