clause-extraction

What problem does it solve?

Manually extracting key contract clauses from PDF and DOCX files is time-consuming, error-prone, and lacks auditable citations for downstream legal operations like CLM backfill, deal diligence, and clause library building.

Core Features & Use Cases

• Citation-grounded clause extraction: Pulls predefined contract clauses (governing law, liability cap, term, indemnification, etc.) from PDF and DOCX files with page and character span citations for full verifiability.
• Legal ops use cases: Supports CLM metadata backfill for legacy contract repositories, pre-deal diligence to surface high-risk terms like change-of-control and liability clauses, and building centralized clause libraries that reflect actual agreed terms across a portfolio.
• Built-in guardrails: Enforces byte-identical excerpt checks to prevent hallucinations, rejects scanned PDFs without OCR layers to avoid false "not present" results, restricts execution to approved Tier-A AI endpoints to prevent privilege leaks, and validates output against a pinned JSON schema to avoid downstream pipeline breakage.

Quick Start

Use the clause-extraction skill to extract all required MSA clauses from the uploaded contract 'acme_vendor_agreement.pdf' and generate a validated JSON record with page-level citations for CLM metadata backfill.