case-research

/case-research

Deep legal research across multiple databases. Routes queries to the right tool, verifies citations against primary sources, and produces a research memo with full attribution.

Instructions

1. Load practice profile. Read ~/.claude/plugins/config/legal-engineering/cross-border-wealth/CLAUDE.md for jurisdiction footprint and risk posture context.

2. Understand the query. From $ARGUMENTS, determine:

   • Is this a specific case lookup? (citation or case name provided)
   • Is this a statute/regulation lookup? (section number or statute name provided)
   • Is this a treaty question? (treaty pair or article mentioned)
   • Is this a topical research question? (e.g., "what's the current position on substance requirements in Cayman?")

3. Pre-flight check. Test whether Westlaw Classic is responding. Record the result for the reviewer note.

4. Route to tools. Use the optimal tool for each query type:

   Query Type	Primary Tool	Fallback
   US case law (search)	westlaw_search_cases	westlaw_classic_search → web search
   US case law (full opinion)	westlaw_get_case	westlaw_classic_get_document
   US legislation/bills	westlaw_search_legislation	web search (congress.gov)
   US regulations	westlaw_search_regulations	web search (federalregister.gov)
   CFR section	westlaw_get_cfr_section	web search (ecfr.gov)
   EU law	westlaw_search_eu_law	web search (eur-lex.europa.eu)
   International/comparative	westlaw_classic_search	web search
   Multi-jurisdiction	westlaw_unified_search	parallel web searches
   Tax treaties	web search (tax treaty databases)	model knowledge with flag
   Non-US case law	westlaw_classic_search	web search (jurisdiction-specific databases)

5. For each source found:

   • Record the full citation (case name, court, date, reporter)
   • Verify the citation against the primary source (is this the actual holding, not dicta, not a dissent, not a quoted argument the court rejected?)
   • Check subsequent treatment (has it been appealed, overruled, distinguished, followed?)
   • Tag with provenance ([Westlaw], [CourtListener], [EUR-Lex], [web search — verify], [model knowledge — verify])

6. Produce the research memo. Structure:

   Question presented: [the research question, precisely stated]

   Short answer: [2-3 sentences — the bottom line]

   Analysis: [structured by issue, with citations. Each legal proposition has a source tag. Jurisdiction-specific throughout.]

   Key authorities:

   Authority	Citation	Jurisdiction	Holding	Status	Source
   [case/statute]	[citation]	[jurisdiction]	[one-line holding]	[current/overruled/under appeal]	[provenance tag]

   Open questions: [what remains unresolved, what needs further research, what the user should verify]

7. Cross-reference with the Star Atlas. If any of the 15 atlas cases are relevant to the research question, note the connection and suggest reviewing the atlas entry.

8. Offer next steps:

   What next?

   1. Go deeper — I'll pull the full opinion(s) and analyze the reasoning in detail
   2. Apply to your structure — I'll apply these findings to [entity/arrangement from practice profile]
   3. Find contrary authority — I'll search for cases that reach the opposite conclusion
   4. HTML report — formatted research memo with linked citations
   5. Send to team — I'll draft a Slack message or email summarizing the findings
   6. Something else

Verification Protocol

• Every citation is verified against at least one primary or authoritative secondary source
• If Westlaw Classic is connected, the [Westlaw] tag requires the citation to appear in a Westlaw result
• Model knowledge citations are ALWAYS tagged [model knowledge — verify]
• The reviewer note lists: tools connected, queries run, results found, verification status
• If a citation cannot be verified, it is flagged: "Could not verify — treat as unconfirmed"