Content
86%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 well-structured skill body: executable commands, bundled scripts that exist and run, four clearly sequenced workflows, and clean one-level-deep reference structure. Weaknesses are minor — a redundant inline Keywords list, a duplicated disclaimer, and the absence of explicit validation/feedback loops in the workflows.
Suggestions
Remove the inline Keywords section (or trim it to terms not already in the frontmatter description) and drop the duplicated footer disclaimer — both duplicate content already carried by the frontmatter and opening paragraph.
Add an explicit validation/feedback step to the workflows, e.g., Workflow 1 step 4: 'Re-run contract_risk_scanner.py on the redlined contract to confirm HIGH findings are resolved before bringing it to counsel', and Workflow 2: 'If the analyzer errors on the JSON, fix the field mapping and re-run'.
Cut common-knowledge one-liners (e.g., copyright arising automatically on creation) or push them into references/ip_and_regulatory.md, keeping only startup-specific deadlines and traps in SKILL.md.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and actionable, but the ~40-term Keywords section duplicates frontmatter triggers, the disclaimer appears twice (intro paragraph and footer), and a few bullets restate knowledge Claude already has ("Copyright — automatic on creation, but register for statutory damages eligibility"). Efficient with minor trims — not 5, since these tokens don't all earn their place; not 3, since there is no padded over-explanation. | 4 / 5 |
Actionability | Quick Start gives copy-paste commands with a bundled-sample fallback ("uses bundled sample if no path given"), both bundled scripts exist and are executable, and the Output Standard is a fill-in template. The single indirection (term-sheet JSON schema via `term_sheet_analyzer.py --help`) is explicitly directed and the schema is documented in the script itself. Not 4: no executable gap remains for the common cases. | 5 / 5 |
Workflow Clarity | Four numbered workflows with clear sequences and some checkpoints ("Always have a securities/venture attorney review before signing", "engage a specialist counsel before building"), but no explicit validate-then-retry feedback loops (e.g., re-run the scanner after redlining, verify the JSON parses before analysis). Matches 'clear sequence with most checkpoints; minor validation gaps' — not 5. No destructive/batch cap applies since operations are read-only analyses. | 4 / 5 |
Progressive Disclosure | SKILL.md is a ~150-line overview with three well-annotated reference files (all of which exist) and two bundled scripts linked from Quick Start; references are one level deep (reference files point only to scripts, no nested md-to-md chains). Clear overview, easy navigation, content appropriately split. | 5 / 5 |
Total | 18 / 20 Passed |