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.
A thorough, well-sequenced workflow with genuinely strong validation checkpoints (backup, fallback chains, review triage, final checklist), but it is significantly over-long: writing-quality rules are stated two to three times, prose philosophy is inlined where a reference file belongs, and one helper code block is pseudocode. Consolidating the duplicated rule sets into a single reference file would improve both conciseness and progressive disclosure at once.
Suggestions
Merge the 'CONFIDENT PROSE, HONEST LIMITS' block, the 'Key Rules' bullets, and the 'Writing Quality Reference' section into one authoritative reference file (or one section) — the same rules are currently stated two to three times.
Replace the comment-only 'Automated bib cleaning' Python block with an actual executable filter script, or move it to scripts/ with the logic implemented.
Move the venue template listings and the 12-rule prose manifesto into references/ files (and ship the templates/ directory the body depends on) so SKILL.md reads as an overview with clearly signaled, one-level-deep references.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is ~420 lines with systematic duplication: De-AI patterns appear in both Step 5 and the 'Writing Quality Reference' section; the 12-rule 'CONFIDENT PROSE, HONEST LIMITS' block is restated as 'Key Rules' bullets ("Calibrate, don't hedge — ... the CONFIDENT PROSE, HONEST LIMITS block above is the contract"), and rules 9-12 get a second summary; plus padding like the Codex/Gemini overlay blockquotes. This is 'noticeably verbose; several unnecessary explanations or padded sections' (anchor 2) rather than the occasional tightening opportunity of anchor 3. | 2 / 5 |
Actionability | Most guidance is executable: venue documentclass snippets, concrete DBLP/CrossRef curl commands ("curl -s https://dblp.org/rec/{key}.bib"), an MCP review prompt with polling instructions, and a 13-item final checklist. Two gaps keep it from a 5: the 'Automated bib cleaning' Python block is comment-only pseudocode with no actual implementation, and the venue templates are referenced via a templates/ directory whose contents are not shown. | 4 / 5 |
Workflow Clarity | Steps 0-8 are clearly sequenced with explicit validation and feedback loops: backup before overwrite ("never silently destroy existing work"), stale-file cleanup, the DBLP→CrossRef→[VERIFY] fallback chain, cross-review with CRITICAL/MAJOR/MINOR severity triage and fix application, a reverse-outline test, and a comprehensive final checklist. This matches the anchor for clear sequence with explicit validation steps, feedback loops, and checklists. | 5 / 5 |
Progressive Disclosure | No bundle files exist (references/, scripts/, assets/ are absent, and the templates/ directory the body depends on is also missing), and the 423-line body inlines content that clearly belongs in separate reference files — the ~40-line prose rules manifesto, the venue template listings, and the section-specific guidelines — while duplicating writing-quality material across two sections. The shared-references links that do exist (output-versioning, output-manifest, output-language) are clearly signaled in a blockquote, and section headers are good, which lifts this above anchor 2's 'minimal structure'. | 3 / 5 |
Total | 14 / 20 Passed |