Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Uses Cowork-style multi-step analysis to process deal room documents -- contracts, financials, org charts. Outputs risk assessment, term comparison, key findings, and negotiation recommendations.
.claude/skills/onewave-ai-cowork-deal-room/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 91% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 577% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 224% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 91% | 0% |
Operate as a senior due diligence analyst using Claude Cowork's multi-step collaboration pattern: work through deal room documents systematically as a team of specialists (legal, financial, organizational) and produce a professional-grade due diligence report. Write with the precision that legal counsel, CFOs, VP of Corporate Development, and M&A advisors expect. Substantiate every finding with a specific document reference. Justify every risk rating.
references/phase-procedures.md -- detailed step-by-step procedure for each of the 5 phasesreferences/classification.md -- the 16 document categoriesreferences/term-extraction-checklist.md -- contract terms to extract in Phase 2references/market-standards.md -- middle-market and large-cap benchmark termsreferences/financial-analysis.md -- financial line items and red-flag checklistreferences/risk-scoring.md -- severity/likelihood scales, domains, composite score, mitigationsreferences/output-formats.md -- console output templates for each phase and the completion summaryreferences/final-report-structure.md -- exact structure for deal-room-analysis.mdTake a directory path containing deal room documents (PDFs, .docx, .xlsx/.csv, text files, images). If no directory path is provided, ask for one before proceeding.
Run all five phases sequentially. Do not skip or combine phases. Each phase feeds the next. See references/phase-procedures.md for the full procedure of each phase and references/output-formats.md for the output template.
[PHASE N/5] Starting: [Phase Name] at the start of each phase.[deal-room-directory]/deal-room-analysis.md following references/final-report-structure.md, then print the completion summary.[filename, page X] or [filename, row X].| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-14 | pass→pass | 8,523 | 6,473 | -24% | 1 | 1 | 0% | 1,155 | 2,086 | +81% | 0 | 0 | — |
case-13 | fail→fail | 9,847 | 5,191 | -47% | 1 | 1 | 0% | 1,577 | 1,965 | +25% | 0 | 0 | — |
case-01 | fail→fail | 51,236 | 7,658 | -85% | 1 | 1 | 0% | 8,245 | 1,554 | -81% | 0 | 0 | — |
case-02 | fail→fail | 11,530 | 8,186 | -29% | 1 | 1 | 0% | 355 | 1,284 | +262% | 0 | 0 | — |
case-03 | fail→fail | 16,480 | 4,163 | -75% | 1 | 1 | 0% | 2,715 | 1,569 | -42% | 0 | 0 | — |
case-04 | fail→pass | 5,953 | 5,357 | -10% | 1 | 1 | 0% | 916 | 1,746 | +91% | 0 | 0 | — |
case-05 | fail→fail | 11,733 | 5,389 | -54% | 1 | 1 | 0% | 1,685 | 1,963 | +16% | 0 | 0 | — |
case-06 | fail→pass | 5,940 | 34,556 | +482% | 1 | 1 | 0% | 912 | 6,176 | +577% | 0 | 0 | — |
case-07 | fail→pass | 10,739 | 31,446 | +193% | 1 | 1 | 0% | 1,669 | 5,408 | +224% | 0 | 0 | — |
case-08 | pass→fail | 20,717 | 9,133 | -56% | 1 | 1 | 0% | 2,811 | 1,801 | -36% | 0 | 0 | — |
case-09 | fail→pass | 12,203 | 6,339 | -48% | 1 | 1 | 0% | 1,769 | 2,146 | +21% | 0 | 0 | — |
case-10 | fail→fail | 8,127 | 8,163 | +0% | 1 | 1 | 0% | 1,219 | 2,217 | +82% | 0 | 0 | — |
case-11 | fail→pass | 6,247 | 5,020 | -20% | 1 | 1 | 0% | 961 | 1,836 | +91% | 0 | 0 | — |
case-12 | pass→pass | 14,382 | 18,321 | +27% | 1 | 1 | 0% | 2,257 | 3,584 | +59% | 0 | 0 | — |
case-15 | pass→pass | 8,807 | 11,244 | +28% | 1 | 1 | 0% | 1,355 | 2,679 | +98% | 0 | 0 | — |
case-16 | fail→fail | 5,813 | 5,685 | -2% | 1 | 1 | 0% | 175 | 1,243 | +610% | 0 | 0 | — |
case-17 | fail→fail | 10,071 | 7,037 | -30% | 1 | 1 | 0% | 1,440 | 1,277 | -11% | 0 | 0 | — |
case-18 | pass→pass | 10,462 | 11,437 | +9% | 1 | 1 | 0% | 1,641 | 2,662 | +62% | 0 | 0 | — |
case-19 | pass→pass | 13,874 | 13,599 | -2% | 1 | 1 | 0% | 2,404 | 3,269 | +36% | 0 | 0 | — |
case-20 | fail→fail | 5,295 | 58,592 | +1007% | 1 | 1 | 0% | 305 | 1,331 | +336% | 0 | 0 | — |
case-21 | pass→pass | 27,857 | 23,355 | -16% | 1 | 1 | 0% | 3,111 | 5,539 | +78% | 0 | 0 | — |
case-22 | pass→pass | 14,001 | 13,393 | -4% | 1 | 1 | 0% | 2,658 | 3,558 | +34% | 0 | 0 | — |
case-23 | pass→pass | 11,986 | 11,277 | -6% | 1 | 1 | 0% | 1,919 | 2,901 | +51% | 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 18 counted toward the lift figure. The other 5 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 +17 percentage points is the difference between those two pass rates over the 18 comparable cases. 1 case got worse with the skill loaded, and it is 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.