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 a well-organized, token-efficient overview with concrete tool-call guardrails and a clear sequenced workflow. Its critical weakness is that every progressive-disclosure pointer (reference/ guides, templates, examples/) targets files that are absent from the bundle, so the promised detail layer is missing entirely.
Suggestions
Add the referenced files to the bundle (reference/citations.md, reference/format-selection-guide.md, reference/advanced-search.md, the four templates, and the examples/ walkthroughs) or remove those references — currently every pointer is dangling.
Inline a minimal example of a Notion:search query and a notion-create-pages payload so the gather/create steps remain executable even if the template files are absent.
Consolidate the Quick start and Workflow sections, which repeat the same five steps, to reduce duplication and reclaim tokens.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and information-dense — no padding explaining what Notion is, and the guardrails convey non-obvious schema facts (e.g., 'The current deployed schema expects both top-level fields even when one is unused'). The main trimmable redundancy is the Quick start section previewing the same five steps as the Workflow section. Not 5 because of that duplication; not 3 because there is no unnecessary concept explanation and nearly every line adds operational value. | 4 / 5 |
Actionability | Concrete tool names throughout (`Notion:search`, `Notion:fetch`, `Notion:notion-create-pages`, `Notion:notion-update-page`) with specific parameter guidance ('include `filters: {}`', 'explicit `parent` and a `pages` array', '`update_properties` with `content_updates: []`') and explicit error handling for missing tools. Not 5 because steps 3-4 depend on templates in `reference/` that are not in the bundle, and no example search query or create payload is given inline. Not 3 because the guidance given is executable, not pseudocode or high-level hints. | 4 / 5 |
Workflow Clarity | Steps 0-5 are clearly sequenced with checkpoints (confirm scope with the user, 'ask the user to confirm if multiple results appear'), plus a recovery path for unavailable tools (pause, connect app, retry). Not 5 because there is no final verification step (e.g., checking the created page or that citations resolved). Not 3 because the checkpoints are explicit, not implicit, and the tool-not-found fallback is a real feedback loop. | 4 / 5 |
Progressive Disclosure | The body points to `reference/` files (citations.md, format-selection-guide.md, advanced-search.md, four templates) and `examples/` walkthroughs, but neither directory exists in the bundle — only `assets/` images (which are unreferenced) are present. Per the guideline to score against the actual bundle structure, the split between overview and detail is broken: the referenced detail layer is entirely missing, so navigation fails at every pointer. Not 3 because the in-body signaling is good and would score well if the files existed; the failure is structural, leaving the skill effectively without its referenced detail files. | 2 / 5 |
Total | 14 / 20 Passed |