get_citator_info
ChatGPTGet citation analysis for a case. Cite each linked authority with its own citation_markdown, including tables and follow-ups. Use heading_markdown when an authority starts a bullet or section. Disclose unavailable links; never use the target case's URL for a citing or cited authority. Returns cases cited by this case, cases citing it, and Paxton's stored status assessment. A status state of "not_assessed" means no stored assessment exists, not good law. Neither an assessment nor citation relationships establish exhaustive subsequent-history review. Args: parent_id: Paxton's case parent identifier. info_type: "cited_by" returns outbound references cited by this case; "citing" returns inbound cases that cite this case; "status" returns the upstream legal-status result; "all" (default) includes all three. Citation counts describe the returned direction.
get_source_text
ChatGPTRetrieve the full text of a returned legal authority. Cite this authority with the returned citation_markdown. Use heading_markdown for a linked heading. Disclose unavailable links. Accepts an administrative source_handle or a statute/case-law s3_path or s3_dir identifier. Returns JSON containing complete normalized text plus resolved identity and citation-link metadata. Mentions of other authorities inside that text do not establish their identity or source URLs. Exactly one source identifier is required. Args: source_handle: Opaque source_handle returned from administrative-decision results. s3_path: Existing s3_path or s3_dir returned from statute or case-law results.
search_administrative_decisions
ChatGPTSearch full EEOC and NLRB administrative decisions with BM25 and sentence windows. Cite authorities with the returned citation_markdown, including short answers and tables. Use heading_markdown when an authority starts a bullet or section. If a link is unavailable, disclose that; never invent one or reuse it for another authority. Covers labor-and-employment decisions from the EEOC and NLRB. Ranking uses full-document BM25 with identity-aware matching; results contain verbatim sentence windows, not generated legal conclusions. Passages can include holdings, allegations, arguments, and quoted authority. Args: query: Keywords, a known identifier or citation, or a double-quoted exact phrase for BM25. research_goal: The user's complete research request, used to rank passages within each decision. agencies: One or both of "eeoc" and "nlrb". Defaults to both. n_results: Number of full decisions to inspect (default and maximum 5). Returns: JSON list of decisions with stable identity, source URL, rank, and verbatim sentence windows carrying exact character offsets into the indexed normalized text.
search_case_law
ChatGPTSearch US case law — federal and state court opinions. Cite authorities with the returned citation_markdown, including short answers and tables. Use heading_markdown when a case starts a bullet or section. If citator_check is present, treat unavailable as unchecked; do not infer good law from a failed status lookup. If a link is unavailable, disclose that; never invent one or reuse it for another authority. Searches Paxton's database of case law using hybrid semantic + keyword search. Args: query: Natural language search query (e.g., "qualified immunity police excessive force") jurisdiction_codes: Filter by jurisdiction (e.g., ["U.S.", "Cal.", "N.Y."]) date_start: Filter cases decided after this date (ISO format, e.g., "2020-01-01") date_end: Filter cases decided before this date (ISO format, e.g., "2024-12-31") include_unpublished: Include unpublished/non-precedential opinions (default True) n_results: Number of matching passages (default 10, max 20); multiple can share one opinion. Returns: JSON list of case law search results with citation, title, text, jurisdiction_code, s3_path, and source_url. The source_url opens the exact opinion in Paxton's authenticated viewer; s3_path is an internal source identifier, not a public URL. Each result's identity describes that opinion, not other authorities mentioned within its text.
search_legal_research
ChatGPTSearch US statutes, regulations, and rules. Cite authorities with the returned citation_markdown, including short answers and tables. Use heading_markdown when an authority starts a bullet or section. If a link is unavailable, disclose that; never invent one or reuse it for another authority. Searches Paxton's database of federal and state laws, regulations, and rules using hybrid semantic + keyword search. Args: query: Natural language search query (e.g., "California dog bite liability") jurisdiction_codes: Filter by jurisdiction slugs (e.g., ["california-codes", "united-states-code"]). Common codes: united-states-code, california-codes, new-york-codes, texas-codes, florida-codes, code-of-federal-regulations, delaware-codes law_type: Filter by law type (statute, regulation, rule). Currently unused, reserved for future. n_results: Number of matching passages (default 10, max 20); multiple can share one authority. Returns: JSON list of search results with citation, legislation_name, text, jurisdiction_code, law_type, slug, and source_url. The source_url opens Paxton's authenticated source viewer; s3_dir is an internal source identifier, not a public URL.
get_citator_info
Claudeget_source_text
Claudesearch_administrative_decisions
Claudesearch_case_law
Claudesearch_legal_research
Claude
