Content
62%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 an exemplary lean router: highly token-efficient with well-structured one-level-deep references. But as a working skill it under-delivers operationally — no workflow sequencing, no validation checkpoints for its batch/destructive-adjacent operations (CLI, link repair), and no inline quick-start, leaving an agent to reconstruct procedure from reference files.
Suggestions
Add a minimal quick-start with one concrete executable example (e.g. a real Obsidian CLI command or a canonical canvas/link-repair snippet) so common cases don't require loading a reference file.
Add a short sequenced workflow with a validation checkpoint for the riskier operations — canvas generation/validation and link repair — e.g. generate, validate, and only then write to `Maps/`.
Either expand the LINK-REPAIR.md and REGISTRY-TABLES.md stubs to hold their actual procedures, or fold their one-line content inline and drop the files, since a one-sentence reference adds navigation cost without payload.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is ~30 lean lines with zero padding: "This skill handles format and artifact concerns only" plus terse bullets and a reference list. It assumes Claude's competence and explains nothing Claude already knows — every token earns its place, matching anchor 5. | 5 / 5 |
Actionability | The Rules are concrete constraints ("`.base` is explicit-only", "literature canvas edges must preserve evidence strength when the source notes provide it") but the body contains no commands, examples, or quick-start — all execution is deferred to references, and two of them (LINK-REPAIR.md, REGISTRY-TABLES.md) are one-sentence stubs, leaving some concrete guidance incomplete per anchor 3. | 3 / 5 |
Workflow Clarity | No sequence exists for any of the eight responsibilities (including batch-like CLI operations and link repair); the body offers only a rough topic map (Responsibilities / Rules / Read next) with no steps and no validation checkpoints, matching anchor 2. It is above anchor 1 because the topic grouping and per-topic references give a rough order of engagement. | 2 / 5 |
Progressive Disclosure | The overview cleanly splits into six real, one-level-deep references (all paths verified to exist) clearly listed under "Read next" and mapped to the Responsibilities section; the minor gap is that LINK-REPAIR.md and REGISTRY-TABLES.md carry less content than the body's own bullets on those topics, so the split is not fully appropriate — anchor 4 rather than 5. | 4 / 5 |
Total | 14 / 20 Passed |