deepxiv/deepxiv_sdk/deepxiv-fa-uk-eu-law

DeepXiv · UK & EU Law (uk_eu_law)

About 52k UK and EU legal instruments (synced daily; exact counts in corpus): about 42.4k EU (EUR-Lex — REG / REG_IMPL / DEC / DIR) plus about 9.3k UK (legislation.gov.uk — ukpga / asp / nia / asc / nisr / ssi / ukla; scanned-PDF-only items and pre-1837 items without text are excluded). Norm types (the same closed set as us_law), regulated parties, penalties, defined terms, deadlines, cross-references and bilingual summaries are model-extracted. Hybrid vector + lexical search; filters and sorting in Postgres.

Corpus: 51,737 documents.

Setup

pip install -U deepxiv-sdk      # deepxiv registers a token on first use

With a registered key (https://data.rag.ac.cn/register), search and read are fair use: 60/min, 5 at once, 5,000/day; on the auto-registered token they spend the deepxiv daily limit at the credits listed below. ask draws on the daily ask allowance (resets 00:00 UTC; a site-wide daily total can run out early at peak times), or 3 lifetime trial asks on the auto-registered token. On 429, back off for Retry-After seconds. Data goes to stdout; cost, quota and sources go to stderr. Exit 0 ok, 1 API error, 2 usage error. The general CLI guide is the deepxiv-cli skill.

Start here

deepxiv fa search uk_eu_law "data protection" -F jurisdiction=eu -k 5
deepxiv fa search uk_eu_law "artificial intelligence high-risk systems" -F kind=REG -F "year>=2020" -k 5
deepxiv fa read uk_eu_law eu:32016R0679 -l brief   # GDPR
deepxiv fa ask "Under the GDPR, when may personal data be transferred to a third country?" -d uk_eu_law

Free samples: ids eu:32016R0679; queries data protection.

Search

deepxiv fa search uk_eu_law "query" [-F FILTER]... [-k N] [--offset N] [--mode M] [--sort S] [--head] [--json]

Modes: hybrid, dense, bm25, keyword. Up to 200 hits. Cost 2 (with --head: 3). Sorts: relevance, date (default relevance). An empty query is rejected; use "*" to list by filters only. Filters combine with AND; an undeclared field exits 2.

Advanced filters (see deepxiv fa spec uk_eu_law): title.

Read

deepxiv fa read uk_eu_law ID [ID...] [-l brief|detail|full] [-e EXTRA] [-p k=v]... [-f json|md|html]

At -l full, -p section=NAME reads one section and -p characters=N the first N characters.

Facets

deepxiv fa facets uk_eu_law FIELD lists the legal values and counts, free. Use it instead of guessing a spelling. Fields: jurisdiction, kind, status.

Ask

deepxiv fa ask "question" -d uk_eu_law [-e low|medium|high] [-p k=v]... [--no-stream] [--json]

Agentic Q&A over this domain: hybrid vector + lexical search (filters the vector layer lacks fall back to Postgres full text); the model reads the hits and answers with [uk_eu_law:<doc_id> <title>] citations; /stream is NDJSON with real token deltas

Always pass -d uk_eu_law: automatic domain picking is still early.

What this corpus will and will not tell you

• Division of labour with us_law: us_law holds US provisions in force (CFR / USC, clause level); uk_eu_law holds UK statutes and EU instruments (whole-instrument level). All three jurisdictions share one closed norm_types set, so 'find every data-protection instrument that carries penalties' is one question across domains.
• The full text is not in Postgres: path in level=detail points at the original on NAS, and source_url is the official link (EUR-Lex / legislation.gov.uk). So this domain has no level=full — fetch the text from source_url.
• summary_zh exists for about 92%; the other 8% have only the English summary. Both are model-written — not quotes.
• status is meaningful only for UK law (set when the title carries '(repealed)'); EU instruments are always empty, so do not filter EU material on status — it would exclude everything.
• subjects is an open model-generated label set and is not filterable before normalisation; filter on norm_types instead.
• Display requires attribution (see the attribution field): © European Union for the EU part, Open Government Licence v3.0 for the UK part.
• Two search paths: with a query, no sort and only filters the vector layer has (not date / title), search is hybrid vector + lexical (semantic recall, terms need not all match), topped up from Postgres; otherwise (sort given, date / title used, or mode=keyword) it runs in Postgres as an English tsvector and a Chinese zhparser pass whose results are unioned — quoted phrases / OR / -exclusion work, and terms are ANDed, so prefer fewer terms plus filters. Each hit's layer says which layer it came from.
• Model-written fields (summarise, do not quote): cross_references, deadlines, defined_terms, norm_types, penalties, regulated_parties, subjects, summary, summary_zh.
• The field table and enumerations are generated from the store's own self-description (search.field), not hand-written; enumeration counts are as of 2026-10-15.

Citing

Cite records as [CELEX 32016R0679]. Say plainly when the corpus has nothing.

Generated by scripts/gen_fa_skills.py from deepxiv fa spec uk_eu_law --json on 2026-10-19. Field names and prices can change; deepxiv fa spec uk_eu_law is the live source.