skills/casedev/redline/SKILL.md
Compares two document versions and produces AI-powered redline analysis with structured change summaries, risk flags, and clause-by-clause breakdowns. Use when the user mentions "redline", "compare documents", "what changed", "version comparison", "contract diff", "track changes", or needs to understand differences between document versions.
npx skillsauth add casemark/skills redlineInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
AI-powered document comparison that goes beyond visual diffs. Upload two versions of a contract or legal document, get a structured analysis of what changed, what it means, and what to watch out for.
Requires the casedev CLI. See setup skill for installation and auth.
# Compare two local files
casedev redline compare ./contract-v1.pdf ./contract-v2.docx --json
# Compare two vault objects
casedev redline compare \
--vault VAULT_ID \
--base OBJECT_ID_V1 \
--target OBJECT_ID_V2 \
--json
casedev redline compare ./old-version.pdf ./new-version.docx --json
Accepts PDF, DOCX, DOC, and TXT. Mixed formats supported (e.g., PDF vs DOCX).
casedev redline compare \
--vault VAULT_ID \
--base OBJECT_ID \
--target OBJECT_ID \
--json
casedev redline compare \
--base-url "https://example.com/v1.pdf" \
--target-url "https://example.com/v2.pdf" \
--json
The default JSON output contains:
{
"summary": "Plain-English narrative of key changes",
"riskFlags": [
{
"severity": "high",
"clause": "Indemnification (Section 8.2)",
"change": "Cap reduced from $5M to $1M",
"impact": "Significantly increases exposure for indemnifying party"
}
],
"changes": [
{
"section": "Section 3.1 — Payment Terms",
"type": "modified",
"before": "Net 30 from invoice date",
"after": "Net 60 from invoice date",
"significance": "medium"
}
],
"stats": {
"sectionsModified": 12,
"sectionsAdded": 2,
"sectionsRemoved": 1,
"riskFlagsHigh": 1,
"riskFlagsMedium": 3
}
}
| Flag | Description |
|------|-------------|
| --format | Output format: json (default), markdown, text |
| --detail | Detail level: summary, standard (default), detailed |
| --focus | Comma-separated clause types to prioritize (e.g., indemnification,termination,liability) |
| --perspective | Analysis perspective: neutral (default), drafter, reviewer |
| --json | Shorthand for --format json |
summary — Executive overview only. Key changes and high-severity risk flags. Fast.standard — Full clause-by-clause breakdown with all risk flags. Default.detailed — Everything in standard plus original/modified text excerpts, suggested responses, and negotiation context.neutral — Objective analysis of changes.drafter — Flags changes that weaken your position (you drafted the original).reviewer — Flags changes you should push back on (you're reviewing incoming edits).A partner sends back a marked-up contract. See what changed and whether to worry:
casedev redline compare ./our-draft.docx ./their-markup.docx \
--perspective reviewer \
--detail summary \
--json
Focus analysis on the clauses that matter most:
casedev redline compare ./v3.pdf ./v4.pdf \
--focus "indemnification,limitation-of-liability,termination,governing-law" \
--detail detailed \
--json
Compare two versions already stored in a vault:
casedev redline compare \
--vault VAULT_ID \
--base OBJECT_ID_V1 \
--target OBJECT_ID_V2 \
--format markdown
Compare across multiple rounds to see the negotiation arc:
# Compare each consecutive pair
casedev redline compare --vault VAULT_ID --base V1_ID --target V2_ID --json > round1.json
casedev redline compare --vault VAULT_ID --base V2_ID --target V3_ID --json > round2.json
casedev redline compare --vault VAULT_ID --base V3_ID --target V4_ID --json > round3.json
| Format | Input | Notes | |--------|-------|-------| | PDF | ✅ | OCR applied automatically for scanned documents | | DOCX | ✅ | Preserves section/heading structure | | DOC | ✅ | Converted to DOCX internally | | TXT | ✅ | Plain text comparison |
Cross-format comparison is supported (e.g., compare a PDF against a DOCX).
"Unable to extract text": Document may be a scanned image without OCR text layer. Re-upload to a vault (automatic OCR) or use casedev ocr process first.
Large documents timeout: Documents over 200 pages may need --detail summary for faster processing.
"No significant changes found": Documents may be identical or near-identical. Check you're comparing the right versions.
Risk flags seem off: Use --perspective to align analysis with your role in the negotiation. Default neutral perspective doesn't assume a side.
tools
Audits the complete in-scope medical-record universe in a litigation matter and produces an attorney-facing, Bates-cited analysis of treatment gaps, missing records or providers, baseline coverage, material billing or production mismatches, and complaint evolution. Use when asked to find missing medical records, analyze treatment gaps or first-care timing, identify absent providers, assess whether a production is complete, or prepare a records-request target list. Use medical-record-chronology instead when the primary request is a chronological clinical narrative.
development
Drafts a legally compliant Private Placement Memorandum for Regulation D offerings (Rule 506(b)/506(c)), covering full disclosure framework including risk factors, capitalization, securities terms, use of proceeds, and investor qualification requirements. Enforces SEC anti-fraud compliance under Section 10(b)/Rule 10b-5, blue sky law considerations, and accredited investor verification under Rule 501. Use this skill when drafting PPMs, offering memorandums, Reg D disclosure documents, or private offering circulars for issuers raising capital from sophisticated investors. Also trigger when the user mentions private placement disclosure, offering memorandum, Reg D fundraising, or accredited investor verification. Even if the user just says "PPM" or "draft our offering memo," use this skill.
data-ai
Generates structured privacy and data protection law briefings across US, EU, UK, and other jurisdictions. Organizes by jurisdiction with compliance deadlines, enforcement actions, and legislative changes. Use when preparing privacy law briefings, compliance updates, regulatory change summaries, or data protection landscape reviews.
testing
Generates structured summaries of prior art references for patent prosecution, validity analysis, and freedom-to-operate assessments. Maps disclosures to claim elements with precise citations. Use when summarizing prior art, analyzing patent landscapes, mapping references to claims, or preparing office action responses.