Content
92%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 strong, highly operational skill body: exact commands for every operation, a pitfalls section packed with genuine failure modes, explicit verification loops, and clean reference splitting. The only meaningful bloat is the inline venv/bootstrap section in Prerequisites, which mixes environment plumbing and authoring meta-notes into the main body.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Dense operational content throughout — Pitfalls bullets like 'NeedAppearances' and 'Rotated stamp text extraction' carry non-obvious failure modes, and the Quick Reference table is maximally token-efficient. However, the ~25-line venv bootstrap block with justification prose ('it survives container restarts, never touches the Hermes install itself') and authoring notes ('Use this exact block in every document skill') could be trimmed or moved out of the main body. | 4 / 5 |
Actionability | Every helper has an exact, copy-paste-ready CLI line with flags ('pdf_split.py doc.pdf --pages 1-3,7 -o part.pdf [--rotate 90]'), the Quick Reference table maps each task to a command, and even external fallbacks come with full invocations (e.g. the Ghostscript 'gs -dPDFA=2 ...' command). Common cases are covered with no pseudocode. | 5 / 5 |
Workflow Clarity | The 9-step Procedure opens with 'Inspect first' and closes with 'Verify', the Verification section gives explicit per-operation checks ('After form fill: pdf_read.py filled.pdf --fields and compare values'), and the form-design loop has a clear feedback cycle ('Iterate spec → lint → overlay until clean'). Destructive/batch operations (encrypt, flatten) all have validation steps, so no cap applies. | 5 / 5 |
Progressive Disclosure | The body is a genuine overview with well-signaled, one-level-deep references that all exist on disk ('references/ocr-extraction.md', 'references/nano-pdf-editing.md', 'references/forms.md'), and all 15 referenced scripts/ files are present. Bulk detail for OCR, forms, and nano-pdf editing is appropriately split out of SKILL.md, and navigation via the command table and labeled sections is easy. | 5 / 5 |
Total | 19 / 20 Passed |