tabular-review

Extracts typed data points from document batches into cited spreadsheet grids.

Updated Jun 17, 2026
One-click install
npx skills add https://github.com/tk1cntt/PhapChe --skill tabular-review-tk1cntt
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: tabular-review
Source: https://github.com/tk1cntt/PhapChe/tree/main/docs/claude-for-legal-main/corporate-legal/skills/tabular-review
Command: npx skills add https://github.com/tk1cntt/PhapChe --skill tabular-review-tk1cntt

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires openpyxl, and includes references (resource) components.

What problem does it solve? Reviewing hundreds of contracts for the same set of questions (change-of-control, assignment, termination) by hand is slow and inconsistent. This Skill builds a review grid — one row per document, one column per data point — where every cell carries a verbatim source quote and location so a reviewer can verify quickly. ## Core Features & Use Cases - Typed schema builder: Define columns as verbatim, classify, date, duration, currency, number, or free text, saved as a reusable .review-schema.yaml file. - Parallel fan-out with citation enforcement: One sub-agent per document returns value, state (answered / not_present / unclear / needs_review), verbatim quote, and location for every cell. - Normalization and spot-check pass: Re-reads cited locations to catch fabricated or mismatched quotes and inconsistent classifications across documents. - Multi-format output: Writes Excel or Google Sheets with color-coded states, hidden source columns, a Flags work queue, and a self-documenting schema sheet, plus CSV and Markdown fallbacks. - Use Case: Run an M&A diligence pass over 200 target contracts to extract change-of-control, assignment, and MAC clause positions into a single auditable workbook. ## Quick Start Ask the assistant to run a tabular review over your contracts folder extracting change-of-control, assignment, and termination provisions into an Excel workbook.

Frequently Asked Questions about tabular-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review hundreds of contracts for the same clauses?

Define a typed schema of columns (e.g., change of control, assignment, term), run a 3-5 document sample to tune it, then fan out one sub-agent per document. Each cell returns a value, state, verbatim quote, and location, compiled into Excel or Google Sheets.

What output formats does a contract review grid support?

The review writes Excel (.xlsx) or Google Sheets as the primary deliverable, plus CSV with a companion _sources.csv and a Markdown table always. Spreadsheet outputs include color-coded states, hidden source columns, a Flags sheet, and a _schema sheet.

Can I reuse a review schema for new documents?

Yes. The schema is saved as a .review-schema.yaml file that you can edit, extend with new columns, and pass back via the --schema flag for re-runs or incremental additions against new document sets.

How does the review handle clauses that are missing or ambiguous?

Cells use three explicit states instead of blanks: not_present when the clause is absent, unclear when drafting is ambiguous, and needs_review when a human judgment is required. The summary reports counts per column as the verification workload.

What are the limitations of automated contract data extraction?

Every cell is a lead requiring human verification, not a finding. The skill does not produce confidence scores, does not replace reading documents, and flags any cell whose quote cannot be verified character-for-character against the source.