CSlawyer1985/legal-skillhub / skills/authority-verification
authority-verification
VerifiedVerifies legal citations and retrieves source material using the case.dev Legal Research API and CLI. Use when validating a citation, confirming a case exists, pulling source text, finding related authorities, extracting citations from a brief or opinion, or building a filing-ready authority check without branded citator terminology.
| File | Kind | Size |
|---|---|---|
| LICENSE.txt | markdown | 10.5 KB |
| SKILL.md | skill md | 3.8 KB |
| references/CASEDEV-WORKFLOW.md | markdown | 3.1 KB |
| references/EVALS.md | markdown | 1.0 KB |
| references/EXAMPLES.md | markdown | 1.7 KB |
| references/OUTPUT-TEMPLATE.md | markdown | 696 B |
| references/RUNTIME.md | markdown | 1.2 KB |
| scripts/verify_authority.py | script | 3.5 KB |
| scripts/verify_authority.sh | script | 1.3 KB |
| scripts/verify_authority.ts | script | 3.1 KB |
SKILL.md
View on GitHub---
name: authority-verification
language: en
description: >-
Verifies legal citations and retrieves source material using the case.dev
Legal Research API and CLI. Use when validating a citation, confirming a case
exists, pulling source text, finding related authorities, extracting
citations from a brief or opinion, or building a filing-ready authority check
without branded citator terminology.
---
# Authority Verification with case.dev
Use case.dev legal research endpoints to verify citations, retrieve source text, and expand to related authorities. Keep the language generic: "verified citation," "candidate authority," "related authority," and "manual treatment review."
## Quick Start
1. Verify the citation text with `legal.verify()` before citing it.
2. If verification is unclear, search by topic with `legal.find()` or `legal.research()`.
3. Pull the full document text with `legal.fullText()` once you have a verified source URL.
4. Expand outward with `legal.similar()` to find related authorities.
5. Format the result using [references/OUTPUT-TEMPLATE.md](references/OUTPUT-TEMPLATE.md).
Reference guide: [references/CASEDEV-WORKFLOW.md](references/CASEDEV-WORKFLOW.md)
Runtime and setup: [references/RUNTIME.md](references/RUNTIME.md)
Concrete scenarios: [references/EXAMPLES.md](references/EXAMPLES.md)
Lightweight eval set: [references/EVALS.md](references/EVALS.md)
Examples:
- Python: [scripts/verify_authority.py](scripts/verify_authority.py)
- TypeScript: [scripts/verify_authority.ts](scripts/verify_authority.ts)
- CLI: [scripts/verify_authority.sh](scripts/verify_authority.sh)
## Core Workflow
### 1. Verify the citation
Use `legal.verify()` with the citation text exactly as written.
- `verified` means the citation matched a real source in the database.
- `not_found` means likely typo, hallucination, or unsupported citation.
- `multiple_matches` means manual review is required before citation.
### 2. Retrieve the source text
If verification succeeds, use the returned case URL with `legal.fullText()` to inspect the opinion and optionally highlight the issue you care about.
### 3. Expand the authority set
Use `legal.similar()` on the verified source URL to find related authorities.
Use `legal.research()` when you need broader issue-based research with multiple phrasings of the same question.
### 4. Audit a document
If you are checking a brief, opinion, or article URL, use `legal.citationsFromUrl()` to extract cited authorities, then verify important citations individually.
### 5. Record the result
Always record:
- the exact citation checked
- verification status
- authoritative URL
- jurisdiction used for search
- date checked
- any manual review needed before filing
## Guardrails
- Do not say a case was "Shepardized," "KeyCited," or given a proprietary signal.
- Do not describe case.dev results as editorial treatment analysis unless you performed separate manual review.
- Say "verified citation" instead of "good law."
- Say "manual treatment review required" when the workflow does not establish precedential weight by itself.
- Search results are candidates, not filing-ready authorities, until verified.
- Re-run verification immediately before filing.
## Output Standard
Format the final work product using [references/OUTPUT-TEMPLATE.md](references/OUTPUT-TEMPLATE.md).
## Dependencies
Do not assume tools are installed. Check [references/RUNTIME.md](references/RUNTIME.md) before running any script.
## Pitfalls
- Verifying only the case name and not the reporter citation.
- Treating `legal.find()` results as verified authorities.
- Assuming related cases from `legal.similar()` are citing references or direct history.
- Skipping the full-text read after verification.
- Hiding uncertainty when the result is `multiple_matches` or `not_found`.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |