Legalcode

By Fordæmi ehf.

Get New App Alerts

Open in ChatGPT

Open in Claude

Overview

Legalcode connects Claude to source-backed legal research covering Iceland, the wider Nordic region, the entire European Union, and 38 other jurisdictions. Use advanced search to narrow results by jurisdiction, source type, date, citation, court, and more. Search legislation, court decisions, regulations, official guidance, agreements, and parliamentary records. Fetch original source text in Markdown when available. Follow links between laws, cases, citations, legislative history, and implementation. Analyze can count and compare a defined set of sources, show trends over time, and break results down by available details. Every result references the underlying primary source and supporting material. Legalcode works across Claude, Cowork, and Claude Code. Results are optimized for follow-up search, tracing, source retrieval, and citation. Search content is processed in memory and never stored. Your documents stay in Claude. Free accounts include non-patent research. Pro adds patent research, higher usage limits, REST API access, CLI tools, and packaged workflows. Legalcode is research software, not legal advice.

Tools

legalcode_analyze

ChatGPT

Analyze supported corpora with counts, rates, averages, groups, matrices, trends, and actor statistics. Read Discover source_profile first: analyticsStatus and analyticsCapabilities are executable truth for the jurisdiction and source. Use court_case for court-only analysis and case for the broad decision rollup. Analysis may require q or a positive narrowing filter for expensive case aggregates, actor analysis, and high-cardinality matrices. q and a body-text filter target the same text, so provide one. Use measure document_count rather than count. Request sourceRef samples only when the profile explicitly allows them; use Search with matching filters for representative documents otherwise.

legalcode_discover

ChatGPT

Discover the exact Legalcode search contract before querying: compact source-owning jurisdictions and verified municipalities, agreement-party scopes, scoped named sources and sourceCode values, supported fields, bounded facets, prefix-constrained facet values, analytics and relationship readiness, query syntax, costs, and examples. Use mode:"jurisdictions" for compact corpus routing and exact source-profile follow-ups; mode:"agreement_scopes" for treaty/agreement coverage; mode:"source_profile" with jurisdiction and sourceType for named sources, fields, and capabilities; mode:"facets" or "facet_values" for stored filter values; and mode:"help" for workflows. Treat case as the broad decision rollup, court_case as confidently classified courts, and agency_decision as administrative, regulatory, or tribunal decisions. Never guess sourceCode or enum values.

legalcode_fetch

ChatGPT

Inspect one non-patent document using a sourceRef copied exactly from Search or Trace. Use legalcode_patent_fetch for patent retrieval. mode:"metadata" returns document context and law-version status. mode:"excerpt" uses q or sectionRef for a targeted passage; supported structured references include Icelandic articles, Irish sections, EU articles, and Icelandic bill commentary sections. Prefer a returned downloadUrl for complete-document retrieval. mode:"full" is a cursor-paginated fallback only when a download is absent or unavailable; follow nextCursor until truncated is false. For IS public-record parents, publicRecordContext pages participants (including ballots and tenure dates), events, or historical/current relations; follow its returned nextAction. An official publisher URL is provenance metadata, not a text-retrieval instruction.

legalcode_patent_fetch

ChatGPT

Retrieve one exact patent sourceRef from legalcode_patent_search. Choose metadata, excerpt around q, or full available claims/description; follow nextCursor until truncated is false.

legalcode_patent_search

ChatGPT

Search EPO OPS bibliographic records by jurisdiction, q, patentId, applicant, inventor, office, or publicationFrom/To. Up to 20 results per page in provider order. Excludes claims and description.

legalcode_search

ChatGPT

Search ranked primary-source candidates and snippets. For topical research, omit mode and use unquoted source-language keywords; ranking already boosts phrase/proximity where supported. Choose one jurisdiction and discovered sourceTypes: data, law (including regulations, narrowed by actType), case, court_case, agency_decision, guidance, agreement, pre_law, or public_record. Use legalcode_patent_search for patents. Prefer explicit identifier filters such as lawKey, CELEX, ELI, citation, or sourceCode for known instruments. q remains free text and never creates structured filters. Do not send raw SQL, fielded backend syntax, wildcards, or a language filter. Results include sourceRef and may include a signed downloadUrl; use Fetch for metadata or targeted excerpts and follow response nextAction guidance.

legalcode_trace

ChatGPT

Trace deterministic indexed relationships. Families include cases_for_law, laws_for_case, cited_cases, citing_cases, pre_law_for_law, related_pre_law_flows, parliamentary_history_for_law, parliamentary_records_for_pre_law, laws_for_parliamentary_record, pre_law_for_parliamentary_record, proceeding_item_for_speech, speeches_and_votes_for_proceeding_item, proposal_for_resolution, eea_incorporation_for_law, implementing_laws_for_eu_law, and eu_law_for_implementing_law. Start from a sourceRef or an exact law, case, citation, or legal-basis key. Short number/year keys require an explicit jurisdiction. Relationship availability and resolution coverage vary by source; inspect Discover source_profile/help before relying on an edge family. For a returned pre-law flow, Search resultLevel:"document" with flowKey to list its documents.