Content
63%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 thorough, actionable guide with concrete code and a clear step sequence, but it carries redundancy between its decision tables and inlines material that would benefit from being split into referenced files. Tightening the duplicated tables and adding a batch failure-retry loop would lift the weakest dimensions.
Suggestions
Consolidate the Step 3 decision tree and the trailing Figure Type Reference table into one table to remove redundancy and recover tokens.
Add a per-figure failure/retry step in Step 5 (e.g. re-run only failed scripts and re-verify) so the batch operation has an explicit feedback loop.
Move the full paper_plot_style.py configuration and the type×style reference into files under references/ and link to them, replacing inline bulk content and the missing shared-references links.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient and task-relevant, but the Step 3 decision tree and the trailing Figure Type Reference table overlap substantially, and prose like "In practice: ~60% / ~40%" plus repeated rules add padding; not a 4 because the redundancy is noticeable rather than minor. | 3 / 5 |
Actionability | Provides copy-paste-ready Python, a bash run loop, LaTeX snippets, and a concrete MCP review call; not a 5 because some examples hardcode values (e.g. values = [82.3, 85.1...]) which contradicts the skill's own "Do NOT hardcode data" rule. | 4 / 5 |
Workflow Clarity | An explicit 8-step sequence with a Step 5 output-existence check and a Step 8 quality checklist supplies most checkpoints; not a 5 because the batch run loop lacks per-figure failure handling and an explicit fix-and-retry feedback loop. | 4 / 5 |
Progressive Disclosure | Section headers give some structure, but large code blocks, the full style config, and duplicate decision/type tables are all inlined, and links point to shared-references/*.md files that are not present as bundle files; not a 4 because content that belongs in separate files stays inline and references are not clearly resolvable. | 3 / 5 |
Total | 14 / 20 Passed |