Drop a contract, get answers. lawclaw rips through PDFs, spots risky clauses, diffs redlines, checks citations, and searches thousands of discovery docs—locally, so nothing leaves your machine. Built for attorneys and paralegals who bill by the hour and can't waste one

Drop a contract, get answers. lawclaw rips through PDFs, spots risky clauses, diffs redlines, checks citations, and searches thousands of discovery docs—loca...

Security Scan

The skill is internally consistent with its stated purpose: it is an instruction-only, local CLI-based toolkit for processing legal documents and does not request credentials or network access; main caveats are that the provided commands can rename/overwrite files and one referenced tool (wdiff) is not declared in the required binaries list.

Purpose & Capability

Name/description match the contents: SKILL.md contains shell workflows (pdftotext, grep, diff, pandoc, pdfinfo, find, wc) appropriate for local contract review, e-discovery, citation checks and redlines.

Instruction Scope

Instructions run local CLI tools against files (extract, grep, diff, rename). This is expected, but several commands modify or rename files (mv, batch pdftotext conversions, writing logs/diffs) so users should expect side effects. Also the SKILL.md uses 'wdiff' but wdiff is not listed as a required binary/install step.

Install Mechanism

Install spec uses Homebrew formulas (poppler -> pdftotext/pdfinfo and pandoc). These are well-known packages; no arbitrary downloads or code extraction are present.

Credentials

No environment variables, credentials, or config paths are requested — consistent with a local, offline document-processing tool.

Persistence & Privilege

always:false and user-invocable:true (defaults). The skill does not request persistent/system-wide privileges or modify other skills' configs.

Assessment

This skill appears to do what it says: local command-line recipes for contract review. Before you use it, back up your documents and run the suggested commands in a copy/test folder because many commands rename, overwrite, or create files (mv, batch pdftotext, redirecting output to logs/diffs). Confirm you have the required tools installed (brew poppler provides pdftotext/pdfinfo; pandoc via brew) and install wdiff if you need word-level diffs (wdiff is referenced but not declared). Review each command in SKILL.md line-by-line to ensure it won't alter files you care about, and run first on non-sensitive samples. If you work with very sensitive client data, consider running these workflows in an isolated environment (air-gapped or VM) to minimize accidental leakage and verify no telemetry is sent by any installed binaries.

lawclaw

Drop a contract, get answers.

lawclaw tears through legal documents the way you wish your associates would—fast, thorough, and without missing the indemnification clause buried on page 47. It extracts text from PDFs, flags key clauses, generates redline comparisons, validates citations, and searches entire discovery sets for the one sentence that wins your case.

Everything runs locally on your machine. No uploads, no third-party servers, no risk to attorney-client privilege. Just you, your terminal, and a very sharp claw.

Who it's for: Attorneys, paralegals, and legal ops teams doing contract review, litigation support, due diligence, e-discovery, or brief-writing.

What it replaces: Hours of Ctrl+F across dozens of PDFs.

Core Capabilities

Document Analysis

• Extract text from legal PDFs: pdftotext <file.pdf> <output.txt>
• Search for specific clauses: grep -i "indemnification\|liability\|warranty" contract.txt
• Word count for billing: wc -w document.txt
• Extract metadata: pdfinfo <file.pdf>

Contract Clause Extraction Use grep with regex to find common clauses:

• Indemnification: grep -i "indemnif\|hold harmless" contract.txt
• Termination: grep -i "terminat\|cancellation" contract.txt -A 3
• Confidentiality: grep -i "confidential\|proprietary\|NDA" contract.txt -A 3
• Force majeure: grep -i "force majeure\|act of god" contract.txt -A 3
• Jurisdiction: grep -i "jurisdiction\|venue\|governing law" contract.txt -A 2
• Arbitration: grep -i "arbitration\|dispute resolution" contract.txt -A 3
• Non-compete: grep -i "non-compete\|noncompete\|restrictive covenant" contract.txt -A 3
• Assignment: grep -i "assign\|transfer\|delegate" contract.txt -A 2

Redline / Comparison

• Compare two versions: diff -u original.txt revised.txt > redline.diff
• Side-by-side view: diff -y original.txt revised.txt | less
• Word-level diff: wdiff original.txt revised.txt > changes.txt
• Convert Word to text first: pandoc contract.docx -t plain -o contract.txt

Citation and Reference Checking

• Find case citations: grep -E "[0-9]+ [A-Z]\.[A-Za-z0-9.]+ [0-9]+" brief.txt
• Find U.S. Code refs: grep -E "[0-9]+ U\.S\.C\. § [0-9]+" document.txt
• Find CFR refs: grep -E "[0-9]+ C\.F\.R\. § [0-9]+" document.txt
• Extract footnotes: grep -E "\[[0-9]+\]" brief.txt
• Find Bluebook short cites: grep -E "[A-Z][a-z]+, [0-9]+ [A-Z]\." brief.txt

Document Organization

• Find specific file types: find . -name "*.pdf" -type f
• Search within discovery docs: grep -r "key term" ./discovery/
• List recent modifications: find . -name "*.pdf" -mtime -7 -ls
• Batch rename exhibits: for f in *.pdf; do mv "$f" "Exhibit_${f}"; done

Discovery Support

• Count pages in PDF: pdfinfo document.pdf | grep Pages
• Batch convert PDFs: for f in *.pdf; do pdftotext "$f" "${f%.pdf}.txt"; done
• Create production log: ls -lh *.pdf > production_log.txt
• Search Bates numbers: grep -r "PROD[0-9]\{6\}" ./documents/

Due Diligence

• Scan for key terms: grep -i "material adverse\|intellectual property\|pending litigation" diligence/*.txt
• Extract dates: grep -E "[0-9]{1,2}/[0-9]{1,2}/[0-9]{4}" contract.txt
• Find monetary amounts: grep -E "\$[0-9,]+(\.[0-9]{2})?" agreement.txt
• Identify parties: grep -E "WHEREAS|Party|Seller|Buyer|Lessor|Lessee" contract.txt

Deposition / Transcript Analysis

• Search testimony: grep -i "Q\." deposition.txt | grep -i "keyword"
• Extract witness answers: grep "A\." deposition.txt -A 2
• Count pages/lines: wc -l transcript.txt

Privilege Log Support

• Generate file list: find ./documents -type f -exec ls -lh {} \; > privilege_log.csv
• Search for privileged terms: grep -ri "attorney-client\|work product" ./emails/
• Tag privileged docs: grep -rli "attorney-client" ./emails/ > privileged_files.txt

Common Workflows

Contract Review Checklist

1. Extract text: pdftotext contract.pdf contract.txt
2. Scan key clauses:
   • grep -i "limitation of liability" contract.txt
   • grep -i "indemnification" contract.txt
   • grep -i "termination" contract.txt
   • grep -i "confidentiality" contract.txt
   • grep -i "governing law" contract.txt
3. Extract monetary terms: grep -E "\$[0-9,]+" contract.txt
4. Flag deadlines: grep -i "within [0-9]+ days\|business days\|calendar days" contract.txt

Redline Workflow

1. Convert both versions: pandoc original.docx -t plain -o original.txt && pandoc revised.docx -t plain -o revised.txt
2. Generate diff: diff -u original.txt revised.txt > changes.diff
3. Side-by-side: diff -y original.txt revised.txt | less

Discovery Review

1. Extract all PDFs: for pdf in discovery/*.pdf; do pdftotext "$pdf" "${pdf%.pdf}.txt"; done
2. Search across all docs: grep -ri "responsive term" discovery/*.txt
3. Generate document index: ls -lh discovery/*.pdf > document_index.txt

Due Diligence Package Review

1. Extract all docs: find diligence/ -name "*.pdf" -exec pdftotext {} \;
2. Search for red flags: grep -ri "litigation\|breach\|default\|bankruptcy" diligence/*.txt
3. Extract unique dates: grep -Eroh "[0-9]{1,2}/[0-9]{1,2}/[0-9]{4}" diligence/*.txt | sort -u

Brief / Motion Citation Check

1. Extract citations: grep -E "[0-9]+ [A-Z]\.[A-Za-z0-9.]+ [0-9]+" brief.txt
2. Cross-reference ToA: diff <(grep -oE "[0-9]+ [A-Z]\.\S+ [0-9]+" brief.txt | sort -u) <(sort toa.txt)

Notes

• All processing runs locally—no cloud uploads—so client confidentiality is preserved
• Convert proprietary formats (DOCX, DOC) to plain text via pandoc before analysis
• Use wdiff for word-level comparison: install via brew install wdiff
• Combine with gog skill for Google Workspace integration (Drive, Gmail, Sheets)
• Confirm with supervising attorney before automated batch operations on case files
• For OCR on scanned PDFs, install tesseract: brew install tesseract

Security & Ethics Reminders

• Verify all OCR-extracted text for accuracy before relying on it in filings
• Maintain privilege logs when processing communications
• Use redaction tools for PII before document production
• Comply with court e-discovery orders, ESI protocols, and local rules
• Keep audit trails of all document processing for chain-of-custody challenges
• Never transmit client data to external services without written consent