Content
62%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-structured, actionable authoring guide with a clear workflow and explicit validation loops. Its main weaknesses are a lack of executable code examples and a monolithic layout with no progressive disclosure to reference files.
Suggestions
Add at least one minimal but executable TypeScript tool definition (name, description, JSON Schema, handler) so the guidance is copy-paste ready rather than prose-only.
Tighten <core_principle> and <context>: drop the GSD-2 internal file-path enumeration and rhetorical flourishes, keeping only the load-bearing design rules.
Move the eval XML format and project skeleton into reference files under references/ (e.g. EVAL_TEMPLATE.md, SKELETON.md) and link to them from the body to enable progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient and does not explain MCP basics, but <core_principle> and <context> carry preachy or boilerplate phrasing ('write it like documentation for a stranger under time pressure', GSD-2 file-path listing) that could be tightened. It is not 3 because some tokens do not purely earn their place, and not 1 because it avoids concept padding. | 2 / 3 |
Actionability | Provides concrete guidance (verb-noun naming, JSON Schema field requirements, error/pagination phrasings, an eval XML template, a skeleton tree), but for a code-producing skill it offers no executable TypeScript example — only a file tree and prose. It is not 3 because the code is not copy-paste executable, and not 1 because the instructions are specific and directive. | 2 / 3 |
Workflow Clarity | A clear six-step sequence with explicit validation checkpoints: Step 4 runs the Inspector and checks error paths, Step 5 runs evals and iterates ('If the model can't complete them, the server — not the model — needs work'), and a success_criteria checklist closes the loop. | 3 / 3 |
Progressive Disclosure | The body is organized into clear tagged sections (objective, context, core_principle, process, anti_patterns, success_criteria) with no deep nesting, but it is a single monolithic file over 50 lines with no bundle references and inline content (eval format, skeleton) that could be split out. It is not 3 because there are no well-signaled one-level-deep references, and not 1 because organization is clean rather than a wall of text. | 2 / 3 |
Total | 9 / 12 Passed |