cited_by
ClaudeList the judgments that CITE this judgment (incoming citations / 'cited by'), newest first, each with a url to open it. Use after get_judgment to show or follow a case's authority. id = the cited judgment's id; limit caps the list (default 30).
fetch
ClaudeFetch one judgment by its id (from a search result). Returns {id,title,text,url,metadata} where text is the VERBATIM text extracted from the official court PDF (High Court PDFs are fetched live; Supreme Court text comes from pre-extraction with a tar fallback). metadata includes court, citation, neutral_citation, cnr, decision_date, judges, disposition, source_pdf_s3_url, dataset_source. If text can't be extracted, that is stated explicitly in text — it is never fabricated.
get_judgment
ClaudeGet a single judgment by id with full metadata and the verbatim judgment text (same retrieval as fetch, richer metadata). Returns id,title,text,url,note,metadata.
get_legal_document
ClaudeGet ONE Indian statutory/legislative document by doc_id (from search_legal_documents) with its FULL verbatim text + metadata (title, document_type, jurisdiction, issuing_authority, issue_date, summary). The text is the actual legal document; if absent it is stated, never fabricated.
list_courts
ClaudeList the distinct courts in the corpus with judgment counts (Supreme Court + High Courts). Useful for discovering exact court names to pass to search_judgments(court=...).
list_document_types
ClaudeList the document types in the statutory corpus (Act, Rules, Notification, Circular, …) with counts — for discovering exact values to pass to search_legal_documents(document_type=...).
lookup_citation
ClaudeLook up judgments by an exact or normalized citation string (matches neutral_citation or law_report_citation, ignoring punctuation/spacing). Returns matching judgments with metadata.
search
ClaudeFull-text search over ALL ~17M Indian Supreme Court & High Court judgments — the entire corpus is searched and ranked; nothing is excluded from matching. Input a natural-language query or keywords (party names, topic, citation). Returns the top-ranked matches as {results:[{id,title,url}]} plus total_matches (how many judgments matched in total). Pass an id to fetch for the verbatim text. Use total_matches to judge breadth: when it is large, switch to search_judgments and NARROW with filters (court, date range, court_type) to the relevant set; page that focused set with offset to retrieve all of it. The full corpus is always searched, so no relevant case is excluded.
search_judgments
ClaudeAdvanced, COMPREHENSIVE search over ALL ~17M Indian SC + High Court judgments — the full corpus is searched and ranked; nothing is excluded. All params optional: query (free text; for the Supreme Court it also searches the FULL judgment text, not just metadata), court (name substring), judge, party, citation, cnr, date_from / date_to (YYYY-MM-DD on decision_date), disposition, court_type ('sc'|'hc'), limit (default 10, max 50), offset (page offset). SCC-style subject filters (Supreme Court, editorial): case_type (one of civil/criminal/constitutional/tax/service/labour_industrial/company_insolvency/arbitration/family/property/contract_commercial/consumer/banking_finance/intellectual_property/environment/election/motor_accident/land_acquisition/writ_pil; comma-separated to OR), area (area of law, comma-separated), act (exact cleaned statute name e.g. 'Negotiable Instruments Act, 1881', or an 'Act name|section' composite e.g. 'Negotiable Instruments Act, 1881|138'), reportable (true/false), cited_by (a judgment id; restricts results to the judgments that CITE that case — its incoming-citation set, composable with query + all other filters). Returns total_matches, count, offset, full metadata + snippet + doc url per hit, sc_count/hc_count, and *_facets (case_type/area/act/disposition/reportable distributions with distinct-judgment counts) to drive a filter rail. Use total_matches + the facets to gauge breadth: when large, NARROW with filters (court, date, court_type, case_type, area, act, disposition, reportable) to the relevant set; then page with offset. Narrowing keeps the result set complete for the user's question while keeping each query fast.
search_legal_documents
ClaudeSearch Indian STATUTORY / legislative documents — Acts, Rules, Notifications, Circulars, Ordinances, Regulations, Orders (Central + all States), ~36K documents with FULL TEXT. This is the LAW ITSELF — use it alongside the judgment tools (case law). Full-text search over the document body. All params optional: query (free text), document_type (e.g. 'Act','Rules','Notification','Circular'), jurisdiction (e.g. 'Central','State (Maharashtra)'), authority, date_from / date_to (YYYY-MM-DD on issue_date), limit (default 10, max 50), offset (paging). Returns total_matches, count, offset, and per-doc {doc_id,title,document_type,jurisdiction,issuing_authority,issue_date,summary,url}. Use get_legal_document(doc_id) for the full verbatim text; page with offset to retrieve every match; list_document_types to discover document_type values.
suggest
ClaudeTypeahead suggestions for search filters. field='citation' returns distinct neutral/law-report citations starting with q; field='party' returns case titles matching the typed party name. For UI autocomplete.
suggest_cases
ClaudeCase-name typeahead returning matching judgments WITH their ids, ranked by authority (incoming-citation count) so landmark cases surface first. Feeds the search UI's 'Cited by' filter picker: pick a case, then filter to the judgments that cite it. Returns [{id, title, court, cited_by_count}].
