Extracts structured data from large sets of legal documents into tabular format for review, analysis, and reporting. Processes contracts, agreements, correspondence, filings, and other legal documents in bulk — extracting key terms, dates, parties, obligations, risks, and custom fields into organized tables. Use when conducting due diligence document review, bulk contract extraction, compliance audits across document sets, lease portfolio analysis, employment agreement review, regulatory filing review, or any task requiring structured extraction from multiple documents. Trigger keywords: tabular review, bulk extraction, document review table, data room review, batch document analysis, contract extraction, portfolio review, structured extraction, document comparison table.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
A direct command skips the review prompt. Inspect the source before running it.
Extracts structured data from large sets of legal documents into tabular format for review, analysis, and reporting. Processes contracts, agreements, correspondence, filings, and other legal documents in bulk — extracting key terms, dates, parties, obligations, risks, and custom fields into organized tables. Use when conducting due diligence document review, bulk contract extraction, compliance audits across document sets, lease portfolio analysis, employment agreement review, regulatory filing review, or any task requiring structured extraction from multiple documents. Trigger keywords: tabular review, bulk extraction, document review table, data room review, batch document analysis, contract extraction, portfolio review, structured extraction, document comparison table.
Extracts structured data from sets of legal documents into tabular format. Each document becomes a row; user-defined questions become columns. Produces reviewable tables with source citations, then supports analysis across the extracted dataset.
Prerequisites
Document set — the files to analyze (contracts, agreements, correspondence, filings, etc.)
Extraction questions — what to extract from each document (see Column Design below)
Document type — the nature of the documents (e.g., all NDAs, mixed data room, email set)
Review purpose — what the table will be used for (DD report, compliance check, portfolio analysis, case chronology)
Output preferences — format (table, CSV, report narrative), language, level of detail
If extraction questions are not provided, propose a standard column set based on the document type.
Workflow
Phase 1: Column Design
Define extraction columns. Each column is a question asked of every document. Column types:
Type
Description
Example
Verbatim
Extract exact language from the document
"What is the governing law clause?"
Free response
Summarize or interpret
"Summarize the key obligations of the Seller"
Classification
Yes/No or category
"Does this agreement contain a change-of-control provision?"
Columns: Employee, Title, Start date, Term, Base compensation, Bonus/equity, Non-compete scope and duration, Non-solicit, IP assignment, Severance triggers, Change-of-control provisions, Governing law
Analysis focus: Non-compete enforceability by jurisdiction, aggregate severance exposure, key person dependencies, inconsistencies across similar roles
Regulatory Filing Review
Columns: Filing type, Filing date, Filer, Jurisdiction, Status, Key disclosures, Material changes from prior filing, Deficiencies noted, Response deadline
Analysis focus: Compliance gaps, missed deadlines, material disclosure changes, cross-filing consistency
Guidelines
Every extracted value must include a source citation (section, page, or paragraph reference)
Distinguish between "Not found" (searched but absent) and "N/A" (not applicable to this document type)
Do not infer or extrapolate values — extract only what the document explicitly states
When a provision is ambiguous, extract the verbatim language and flag for attorney review rather than interpreting
Process documents in consistent order (alphabetical, chronological, or as provided)
For large sets, process in batches and maintain consistent column definitions across batches
Flag any document that appears to be a duplicate or superseded version
Note document language — if documents are in multiple languages, indicate the source language and whether extraction was from the original or a translation
Mark [VERIFY] on any extracted value where confidence is low due to poor document quality, ambiguous language, or complex cross-references
Maintain a processing log: document name, status (processed/skipped/flagged), notes
