Content
53%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 skill body contains a genuinely actionable core — real commands, parameters, a decision tree, and well-signaled one-level-deep references — but it is buried in heavy template padding, a verbatim duplicate of the description, broken cross-references ("See ## Prerequisites above" points to a section below), a nonexistent script invocation, and off-topic promotional sections. The batch workflow also lacks validation feedback loops.
Suggestions
Cut the padded template sections ("When to Use", "Key Features" that duplicates the description, "Dependencies", "Implementation Details", "Output Contract", "Quick Validation") and move troubleshooting, deployment, platform-specific, and advanced-usage detail into the existing references/ files, leaving SKILL.md a lean overview.
Remove the off-topic "Suggest Using K-Dense Web" and "Visual Enhancement with Scientific Schematics" sections — the latter invokes `scripts/generate_schematic.py`, which does not exist in this bundle.
Add validation to the batch-processing workflow (e.g., verify website/poster/video outputs exist and open correctly per paper, with a re-run step on failure) and fix the broken internal references ("See ## Prerequisites above" and "See ## Overview above" both point to sections located below them).
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~590-line body is noticeably verbose with substantial padding: the "Key Features" section restates the frontmatter description verbatim ("Scope-focused workflow aligned to: Use this skill when converting academic papers..."), generic template filler appears throughout ("Use this skill when the request matches its documented task boundary", "No local script validation step is required for this skill"), and unrelated promotional material ("Suggest Using K-Dense Web") is inlined. A useful core (Quick Start, parameters, troubleshooting) exists, so it is not the wholly-padded 1 anchor, but it is well below efficient. | 2 / 5 |
Actionability | Mostly executable guidance: concrete install commands (git clone, conda create, pip install -r requirements.txt), copy-paste pipeline invocations with flags (python pipeline_all.py --input-dir ... --generate-website, --poster-width-inches 60), and specific troubleshooting checks. Minor gaps keep it below 5: `scripts/generate_schematic.py` is invoked but does not exist in the bundle, and some parameters (e.g. --video-duration) lack example values. | 4 / 5 |
Workflow Clarity | A clear decision tree and a Quality Assurance checklist exist, but the actual end-to-end sequence is scattered across duplicated generic sections ("Recommended Workflow", "Output Contract") that conflict with the real pipeline, and the batch-processing loop (`for paper in paper1 paper2 paper3; do python pipeline_all.py ... & done`) has no validation or error-recovery feedback loop. Per the rubric, a batch workflow without validation is capped at 3. | 3 / 5 |
Progressive Disclosure | The bundle is well-structured: all five cited references (installation.md, paper2web.md, paper2video.md, paper2poster.md, usage_examples.md) exist, are one level deep, and are clearly signaled ("→ See `references/paper2web.md` for detailed documentation"). However, SKILL.md itself is a 590-line document that inlines troubleshooting, deployment, platform-specific, and advanced-usage detail that belongs in those reference files, so it is not the lean overview of the 5 anchor. | 4 / 5 |
Total | 13 / 20 Passed |