Content
56%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 delivers genuinely actionable material — executable commands, parameter tables, formulas, a stage-classification rubric, and a checkpointed workflow with fallbacks — but buries it in roughly a hundred lines of generic boilerplate sections and inlines what should live in reference files. The result is a bloated SKILL.md that dilutes its own good content.
Suggestions
Cut the boilerplate sections that add no task-specific knowledge (Key Features, Implementation Details, Output Requirements, Risk Assessment, Security Checklist, Evaluation Criteria, Lifecycle Status) and fix the dangling "See `## Workflow` above" pointer.
Move the I/O schemas, algorithm description, stage judgment criteria, and configuration tables into a references/ file (e.g. references/algorithm.md), keeping SKILL.md as a concise overview with one-level-deep links.
Verify and correct the Python API import path and replace the opaque `cd "20260318/scientific-skills/..."` example path with the actual relative path from the skill root.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~295-line body carries pervasive padding: "Key Features" restates the description, "Implementation Details", "Output Requirements", "Security Checklist", "Evaluation Criteria", and "Lifecycle Status" are generic boilerplate that adds nothing Claude does not already know, and "See `## Workflow` above for related details" is a dangling self-reference placed before that section exists. It is not 3 because the padding is extensive rather than occasional, and not 1 because genuinely useful task-specific content (parameter tables, formulas, runnable commands) is interleaved with the filler. | 2 / 5 |
Actionability | Concrete, mostly executable guidance is present: `python scripts/main.py --keyword "artificial intelligence" --data-file data.json`, a runnable Python API example with real data, parameter tables with defaults, explicit velocity/acceleration formulas, and a py_compile quick check. It is not 5 because the import path `from skills.keyword_velocity_tracker.scripts.main import ...` is unverified against the actual bundle layout and the usage example cds into an opaque dated path, and not 3 because the commands and code shown are executable rather than pseudocode. | 4 / 5 |
Workflow Clarity | The Workflow section gives a five-step sequence with explicit checkpoints (confirm objective and inputs, validate scope and stop early on unsupported assumptions, run with available inputs, return a structured result, switch to fallback on failure), reinforced by the Quick Check and Audit-Ready Commands sections and an Error Handling section naming failure points. It is not 5 because steps remain high-level (e.g. "choose the packaged workflow") without per-step commands or a fix-and-retry loop, and not 3 because validation checkpoints are explicitly present rather than missing. | 4 / 5 |
Progressive Disclosure | The body is well-sectioned and its one reference (references/audit-reference.md) is real and properly signaled in a References section, but ~150 lines of I/O schema, algorithm description, stage criteria, and configuration detail are inlined in SKILL.md when they belong in references/ — only a thin audit-reference file exists alongside them. It is not 4 because the split between overview and detail is not "mostly appropriate" (the bulk of reference material sits inline), and not 2 because section structure is clear and the existing reference is clearly signaled. | 3 / 5 |
Total | 13 / 20 Passed |