deepxiv/deepxiv_sdk/deepxiv-fa-fr

DeepXiv · Federal Register (fr)

1.0M Federal Register documents (1994– ): rules, proposed rules, notices and presidential documents. Action kind (closed set of 19), issuing agencies, affected parties, effective and comment-close dates and one-line summaries are model-extracted. No vectors: structured filters plus bilingual full-text search in Postgres.

Corpus: 1,008,597 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 fr "PFAS drinking water" -F doc_type=Rule -k 5
deepxiv fa search fr "artificial intelligence" -F doc_type="Proposed Rule" -F "pub_date>=2024-01-01" --sort pub_date -k 5
deepxiv fa read fr fr:2024-07332 -l brief
deepxiv fa ask "What did EPA finalize on PFAS in drinking water?" -d fr

Free samples: ids fr:2024-07332; queries PFAS drinking water.

Search

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

Modes: keyword. Up to 200 hits. Cost 2 (with --head: 3). Sorts: relevance, date, pub_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 fr): norm_tags, cfr_refs, comments_close_on, docket_ids, document_number, effective_on, fr_topics, rins, significant.

Read

deepxiv fa read fr 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 fr FIELD lists the legal values and counts, free. Use it instead of guessing a spelling. Fields: action_kind, doc_type.

Ask

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

Agentic Q&A over this domain: with no vector layer, the model turns the question into structured filters plus bilingual full-text search; the model reads the hits and answers with [fr:<doc_id> <title>] citations; /stream is NDJSON with real token deltas

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

What this corpus will and will not tell you

• Division of labour with us_law: us_law holds the provisions in force (clause-level CFR / USC); fr holds the notices themselves (who changed what, when).
• doc_type uses the official capitalisation (Rule / Proposed Rule / Notice / Presidential Document / Uncategorized Document / Correction / Sunshine Act Document); lowercase 'rule' matches nothing.
• Rules open for comment: filters.comments_close_on={"gte": "<today>"}; upcoming effect: a range on effective_on.
• subjects is an open model-generated label set and is not filterable before normalisation — filter on fr_topics (the official topics) instead.
• No vectors: the query runs as an English tsvector and a Chinese zhparser pass whose results are unioned; mixed queries are split automatically, and quoted phrases / OR / -exclusion work.
• Model-written fields (summarise, do not quote): action_kind, affected, key_point, norm_tags, one_liner, one_liner_zh, subjects.
• 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 [89 FR 12345]. Say plainly when the corpus has nothing.

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