cite-check

Verify every legal citation in a document against the CourtListener database.

72|19|Updated Jun 6, 2026
One-click install
npx skills add https://github.com/sure-scale/doc-haus --skill cite-check-sure-scale
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: cite-check
Source: https://github.com/sure-scale/doc-haus/tree/main/dochaus/skill/cite-check
Command: npx skills add https://github.com/sure-scale/doc-haus --skill cite-check-sure-scale

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI-assisted legal writing frequently produces fabricated or mischaracterized citations, and courts sanction filers for them. This Skill audits every citation in a brief, memo, motion, or opinion letter against a live case-law database so no authority is trusted from memory. ## Core Features & Use Cases - Full Citation Extraction: Lists every case, statute, regulation, and rule cited, resolving short-form references like id. and supra back to their antecedents. - Database Verification: Checks each case against CourtListener (via MCP server or the case-law tool) by name and reporter citation, confirming court, year, and that the holding supports the cited proposition. - Four-Status Classification: Labels each citation as verified, not found, ambiguous, or supports-different-proposition, with a findings table including anchors, evidence URLs, and specific mismatches. - Use Case: Before filing a motion drafted with AI assistance, run a cite-check to catch a hallucinated case citation and avoid sanctions. ## Quick Start Ask the assistant to cite-check the attached brief and verify every citation against CourtListener before filing.

Frequently Asked Questions about cite-check

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

FAQPage Schema
How do I verify legal citations in a brief before filing?

Extract every citation from the document, then look up each case by name and reporter citation in CourtListener. Confirm the court, year, and that the retrieved content supports the proposition cited, and classify each citation as verified, not found, ambiguous, or supports-different-proposition.

How to detect hallucinated case citations in AI-generated legal writing?

Treat every citation as unverified until a live database lookup returns the case. A citation with no match by name or reporter citation in CourtListener is classified as not found and presumptively fabricated until the author produces the source.

Does CourtListener verification work for non-U.S. cases and statutes?

CourtListener covers U.S. opinions only. Non-U.S. authorities, statutes, and regulations cannot be verified through it and are classified as ambiguous, then listed for manual confirmation rather than guessed at.

What happens when a real case is cited for the wrong proposition?

The citation is classified as supports-different-proposition. A real case cited for something it does not say is treated as seriously as a fabricated one, and the specific mismatch is recorded in the findings table.

Can a citation be marked verified from memory without a database lookup?

No. Verification requires a live lookup that returned the case with matching reporter citation, court, and year. A document is safe to file only when every citation is verified or independently confirmed by the supervising attorney.