Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Reads a contract and generates redline suggestions with replacement language. Identifies unfavorable terms, missing protections, ambiguous language, liability exposure, IP risks, termination traps, and auto-renewal gotchas. Produces a contract-review.md with clause-by-clause analysis, risk ratings, tracked changes format, and negotiation talking points. Use when the user wants redline markup, contract markup, or suggested contract edits.
.claude/skills/onewave-ai-contract-redliner/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | 109% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 117% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 88% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 117% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 47% | 0% |
Read a contract and produce a contract-review.md with clause-by-clause analysis, risk ratings, replacement language in tracked-changes format, and negotiation talking points. Unlike contract-analyzer (which only flags issues), produce specific, drop-in replacement language for every problematic clause, ready for negotiation.
references/risk-categories.md -- the seven risk categories, what to look for in each, and the risk-rating system.references/redline-format.md -- per-issue redline entry format and tracked-changes conventions.references/output-template.md -- the full contract-review.md structure to generate.references/contract-types.md -- per-contract-type focus areas and worked examples..txt, .md, .pdf, .docx), or a URL. Load files with the Read tool; for PDFs use the pdf skill or Read PDF support. Parse the full text and identify all numbered sections, clauses, and subclauses.references/contract-types.md.references/risk-categories.md. Assign each section a rating (CRITICAL / HIGH / MEDIUM / LOW / ACCEPTABLE). Mark fair sections as ACCEPTABLE with a brief note -- do not skip them.references/redline-format.md. Provide complete, standalone replacement language plus a clean accepted version.contract-review.md. Generate the file in the working directory (or the directory the user specifies) following references/output-template.md exactly and in order.> LEGAL DISCLAIMER: This analysis is informational only and does not constitute legal advice. Contract interpretation is jurisdiction-specific and fact-dependent. Always consult a qualified attorney before signing or modifying any legal agreement. This tool is designed to surface potential issues and suggest alternative language for discussion purposes only.
contract-review.md as the primary deliverable.[-deletion-] / [+insertion+]) for every suggested change.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 9,255 | 9,630 | +4% | 1 | 1 | 0% | 852 | 1,079 | +27% | 0 | 0 | — |
case-02 | fail→fail | 7,932 | 7,221 | -9% | 1 | 1 | 0% | 409 | 1,105 | +170% | 0 | 0 | — |
case-03 | fail→fail | 4,203 | 6,107 | +45% | 1 | 1 | 0% | 697 | 1,086 | +56% | 0 | 0 | — |
case-04 | pass→fail | 17,392 | 35,954 | +107% | 1 | 1 | 0% | 3,177 | 6,003 | +89% | 0 | 0 | — |
case-05 | fail→fail | 2,356 | 4,088 | +74% | 1 | 1 | 0% | 303 | 1,237 | +308% | 0 | 0 | — |
case-06 | pass→pass | 22,150 | 19,653 | -11% | 1 | 1 | 0% | 3,181 | 4,229 | +33% | 0 | 0 | — |
case-07 | pass→pass | 5,883 | 8,146 | +38% | 1 | 1 | 0% | 805 | 2,115 | +163% | 0 | 0 | — |
case-08 | fail→pass | 11,681 | 21,336 | +83% | 1 | 1 | 0% | 2,201 | 4,606 | +109% | 0 | 0 | — |
case-09 | fail→pass | 9,385 | 43,418 | +363% | 1 | 1 | 0% | 1,689 | 3,659 | +117% | 0 | 0 | — |
case-10 | fail→pass | 7,390 | 24,360 | +230% | 1 | 1 | 0% | 1,105 | 2,079 | +88% | 0 | 0 | — |
case-11 | fail→pass | 11,583 | 18,908 | +63% | 1 | 1 | 0% | 1,867 | 4,044 | +117% | 0 | 0 | — |
case-12 | pass→fail | 16,439 | 14,022 | -15% | 1 | 1 | 0% | 2,611 | 3,259 | +25% | 0 | 0 | — |
case-13 | pass→fail | 10,812 | 18,905 | +75% | 1 | 1 | 0% | 1,796 | 2,511 | +40% | 0 | 0 | — |
case-14 | pass→pass | 14,454 | 21,631 | +50% | 1 | 1 | 0% | 2,375 | 4,633 | +95% | 0 | 0 | — |
case-15 | fail→pass | 17,098 | 16,069 | -6% | 1 | 1 | 0% | 2,361 | 3,482 | +47% | 0 | 0 | — |
case-16 | fail→fail | 14,828 | 14,997 | +1% | 1 | 1 | 0% | 2,475 | 3,511 | +42% | 0 | 0 | — |
case-17 | fail→fail | 4,203 | 38,242 | +810% | 1 | 1 | 0% | 653 | 6,959 | +966% | 0 | 0 | — |
case-18 | pass→pass | 13,136 | 11,877 | -10% | 1 | 1 | 0% | 2,039 | 2,875 | +41% | 0 | 0 | — |
case-19 | pass→pass | 12,496 | 20,945 | +68% | 1 | 1 | 0% | 1,838 | 3,906 | +113% | 0 | 0 | — |
case-20 | pass→pass | 13,502 | 15,372 | +14% | 1 | 1 | 0% | 2,134 | 3,438 | +61% | 0 | 0 | — |
case-21 | pass→pass | 15,053 | 23,254 | +54% | 1 | 1 | 0% | 2,389 | 4,478 | +87% | 0 | 0 | — |
case-22 | pass→pass | 10,960 | 16,859 | +54% | 1 | 1 | 0% | 1,969 | 3,699 | +88% | 0 | 0 | — |
case-23 | pass→pass | 12,086 | 18,952 | +57% | 1 | 1 | 0% | 1,910 | 4,140 | +117% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted, and 20 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +9 percentage points is the difference between those two pass rates over the 20 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.