Search requires at least one jurisdiction, framework, sector, or source; it does not auto-detect scope from the query. Find laws and regulations on a specific topic. TIER LIMITS: free tier takes ONE jurisdiction or ONE framework per call — no sectors=, no multi-source scope, and no fan-out to agency guidance, case law, or preparatory works — with 100 searches/day and 3 concurrent calls. Solo lifts those scope limits — several jurisdictions, frameworks, and sources in one call — at a higher daily budget; the evidence layer stays premium+. Paid tiers carry high daily abuse ceilings (premium 5,000/seat; team 50,000 and company 500,000 pooled per organisation) sized well above legitimate use. Premium+ unlocks multi-jurisdiction server-side fan-out to agency-guidance, case-law, and preparatory-works MCPs alongside primary legislation — all returned in one search response; there is no separate search_case_law or search_preparatory_works tool to call. (For agency guidance only, the standalone search_guidance tool is also available if you want guidance results without primary-law fan-out.) Call get_my_capabilities to inspect your tier and remaining quota before issuing large batches. Use this when someone asks 'what does the law say about X in country Y' or 'which regulations cover Z'. Examples: • 'GDPR data retention rules' → search(query='data retention', frameworks=['GDPR']) • 'Swedish consumer protection' → search(query='konsumentskydd', jurisdictions=['SE']) • Cross-cutting industry vertical → search(query='vehicle cybersecurity', sectors=['automotive']) reaches automotive-cyber MCPs (UN ECE WP.29) regardless of jurisdiction • Topic without a country → call list_coverage first; present jurisdictions to the user or ask which applies • License questions need explicit sources= (no jurisdiction): - Software / code / SPDX / REUSE / CRA / NOTICE templates → sources=['data-use-license'] - Dataset licences / government open-data terms (NLOD, Etalab, OGL, DL-DE, IODL, KOGL) / vendor TOS (Westlaw, LexisNexis, Bloomberg) / Crown Copyright / EU Database Directive / Creative Commons family → sources=['data-use-license'] Always pass at least one of jurisdictions (ISO-2 like 'SE'/'DE'), frameworks ('GDPR', 'NIS2'), sectors ('automotive', 'aviation', 'insurance', 'maritime'), or sources. A scope value that names nothing we serve is REFUSED, not ignored: the whole call errors and names the value (with the nearest known id where there is one), so a mixed scope never returns rows for only the part that resolved — the gateway does not answer a narrower question than the one you asked. frameworks= filters the source set to MCPs that declare framework coverage — the query string is still FTS-matched against each source's native corpus, with no automatic mapping from framework controls to jurisdiction transposition articles. For cross-framework control mapping (e.g. ISO 27001 ↔ SOC 2 ↔ NIST 800-53), pass frameworks=['ISO_27001','SOC_2',...] — server-side fan-out includes Security Controls MCP. sectors= reaches industry-vertical MCPs that span multiple jurisdictions (UN ECE / EASA / IAIS regimes). jurisdictions= combined with sectors= is an INTERSECTION — only the sector's MCPs within those jurisdictions; an empty intersection returns an explicit error naming the jurisdictions that do carry the sector; sectors= alone stays cross-jurisdictional. Search the target language natively (e.g., SE: konsumentskydd, DE: Datenschutz, FR: protection des consommateurs). CJK scripts (Japanese/Chinese): never space-separate a compound — spaced tokens are ANDed and may never co-occur as separate tokens in unsegmented CJK text; pass the compound unspaced (個人情報保護, not 個人情報 保護). QUERY SHAPE: queries are keyword-matched (FTS5, implicit AND — every term must occur in the SAME provision). Pass one or two canonical concept terms per call; never a multi-concept compound. A compound such as 'incident reporting deadline personal data breach' returns 0 even when each concept on its own returns hits — so ask one concept per call and combine the answers yourself. Two terms describing ONE concept ('personal data') are fine; 2-3 alternative terms can be joined with a bare uppercase OR (e.g. 'spoofing OR tampering' matches either term). OR is for synonyms of ONE concept, not for related concepts — 'dismissal OR termination' yes, 'encryption OR breach notification' no (ask those one per call). Other FTS operators (AND, NOT, NEAR) are stripped. STRICT MISS: when a search completes cleanly and no result matched your terms strictly, the response carries meta.outcome = 'NO_STRICT_MATCH'. The recovery fields — meta.recommended_action, meta.recommended_scopes, meta.broadening_available — are set on any qualifying strict miss, INCLUDING a partial fan-out where outcome stays null, so read them whenever present, not only under an outcome. On a partial fan-out meta.broadening_available stays null when the missing leg makes it unknowable — null there means unknown, never 'no'. On recommended_action = 'RETRY_ONE_CONCEPT_PER_CALL', re-issue the search with ONE concept per call. On recommended_action = 'OFFER_BROADENING_TO_USER' — and wherever meta.broadening_available is true — relaxed matches exist and are withheld: tell the user, offer a re-run with allow_broadening=true (served rows are stamped match_mode='broadened' and pass the same relevance floor), and re-run only if the user accepts — never broaden on your own. meta.recommended_scopes names scope ids that were not searched. If 0 results, tell the user; do not answer from training data. The response ends with a 'Sources used' section — a markdown table carrying the audit receipt for each returned row, or a labelled zero-result note — and meta.render_contract carries the versioned evidence-curation contract for reproducing source attributions when the answer is rendered.