Content
68%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 exceptionally concise and well-structured with concrete, tool-specific guidance for the Sanity/Linear workflow. Its two real weaknesses are the missing reference.md bundle file — which the content leans on for all templates and IDs — and the absence of validation/verification checkpoints for document creation and other batch/destructive operations.
Suggestions
Ship the referenced reference.md in the bundle (or inline the essential Portable Text template, the GROQ query for reference IDs, and the changeBlock template into SKILL.md) — currently every link in the body points to a nonexistent file.
Add a verification step after create_documents/patch_documents (e.g., re-fetch the draft by _id and confirm the content array is intact) so the workflow has a validate→fix→retry loop for its batch document operations.
Include an explicit post-create error-recovery note for the most common failure the skill itself identifies (invalid JSON in create_documents), such as re-checking brace/bracket balance and retrying before reporting failure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes competence: it never explains what Sanity, Linear, Portable Text, or MCP are, and every line carries operational knowledge ("Every block/span needs a unique `_key`", "brace/bracket errors are the most common failure", "MCP cannot upload local files"). It matches anchor 5 — every token earns its place — and is well above anchor 4's 'minor instances of over-explanation'. | 5 / 5 |
Actionability | Concrete guidance throughout: named tools (`create_documents`, `patch_documents` with `insert` `before`/`after`/`replace`), an executable patch-target example (`content[_key=="fd1ce739c172"]`), runtime ID resolution via GROQ, and specific rules (draft-only, unset `publishedAt`). It falls short of anchor 5 because the actual Portable Text templates and GROQ queries are deferred to reference.md — and reference.md is absent from the bundle, so the copy-paste-ready material is not actually reachable. | 4 / 5 |
Workflow Clarity | The numbered checklist (steps 1–7) gives a clear sequence with useful guards ("confirm intent" before the permanent `discard_drafts`, publish only when asked), but there is no validation or verification step after `create_documents` and no validate→fix→retry loop for a workflow built on document/database operations. Per the guideline capping workflow clarity at 3 for destructive/batch operations without validation, this lands on the anchor-3 boundary rather than 4. | 3 / 5 |
Progressive Disclosure | The body is itself well-sectioned and consistently points one level deep to [reference.md](reference.md) with anchors (#fetch-reference-ids, #portable-text-cheat-sheet, #changeblock-template), but reference.md does not exist — there is no references/, scripts/, or assets/ directory in the bundle. Every template, GROQ query, and connection constant the body depends on is therefore unreachable, which is worse than inlining and matches anchor 2's 'minimal structure / broken organization' territory rather than anchor 4's 'references mostly clear'. | 2 / 5 |
Total | 14 / 20 Passed |