Content
75%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 a lean, well-organized meta-skill whose core instruction — never trust memory, verify against the version-matched bundled docs — is reinforced with concrete commands, exact paths, and a type-check validation loop. It scores a solid 4 across all dimensions: consistently efficient and actionable, with only minor trimming, sequencing, and splitting opportunities keeping it from the top anchors.
Suggestions
Trim descriptive sentences that restate what a tool does (e.g. the DevTools capture capabilities sentence) to pure instructions, moving conciseness toward the 'every token earns its place' anchor.
Add one small copy-paste-ready snippet for the most common case (e.g. a minimal generateText or useChat example marked as 'verify against bundled docs before use') so actionability covers common cases, not just verification commands.
Present the core procedure as an explicit numbered sequence (check installed version → read bundled docs → implement → type-check → fix from docs) so the workflow's validation checkpoints are as explicit as its topic sections.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is directive and largely free of padding — e.g. 'Never write AI SDK code from memory. Always verify every API... against the documentation and source code' — with only minor over-explanation such as the DevTools summary ('captures your AI SDK calls - requests, responses, tool calls, token usage, and multi-step runs'). This matches anchor 4 ('efficient; minor instances of over-explanation') better than anchor 5, since a few sentences could still be trimmed. | 4 / 5 |
Actionability | It gives concrete, runnable commands ('pnpm add ai', 'curl -s https://ai-gateway.vercel.sh/v1/models | jq -r ".data[].id"', 'npm view ai version') and exact file paths ('node_modules/ai/docs/', 'node_modules/ai/src/'), but contains no executable AI SDK code examples, deferring all code to the docs. This is 'mostly executable guidance with minor gaps' (anchor 4) rather than anchor 5's copy-paste-ready examples covering common cases. | 4 / 5 |
Workflow Clarity | The workflow is clear and ends with an explicit validation/feedback checkpoint ('Run the project's type checker... Most type errors come from remembered, now-changed APIs; re-check the current docs and source when they occur'). It matches anchor 4 ('clear sequence with most checkpoints present; minor validation gaps') — the sections are organized by topic rather than a strictly numbered sequence, and no destructive/batch cap applies. | 4 / 5 |
Progressive Disclosure | No bundle files exist, and the body (~74 lines) exceeds the under-50-line simple-skill exception, so this is scored on structure alone. Sections are well-organized and deferral of detail is clearly signaled ('read the bundled guide and the AI Gateway docs', 'For setup instructions, read the bundled DevTools documentation'), matching anchor 4's 'good structure; most content appropriately placed; minor organization gaps' — e.g. the Gateway setup and DevTools sections could be split out were the skill to grow. | 4 / 5 |
Total | 16 / 20 Passed |