Federal Regulations Mcp Server

Overview

US federal regulations from the Federal Register, the eCFR, and Regulations.gov. Search proposed and final rules, trace a rule from proposal through public comments to codified text, and read any CFR section as it stands today or stood on a past date.

Tools

Resources

The two data resources mirror regulations_get_document and regulations_get_cfr_section. The federal-regulations guide also ships with plugin installs, npm, Docker, and the desktop bundle.

Capability reference

regulations_search_rules _tool

• Optional full-text query, filtered by type (PRORULE / RULE / NOTICE / PRESDOCU), agencies (Federal Register slugs such as environmental-protection-agency, not names or acronyms), published_after / published_before, cfr_title + cfr_part (a part number or range such as 140-143; cfr_part needs cfr_title, else title_required_for_part), docket_id (a docket number as the Federal Register prints it), rin, significant (the Federal Register's EO 12866 flag: true and false each match only documents with a determination), and effective_after / effective_before, all combined; per_page from 2 (default 20; above 100 serves 100 with a notice), page 1–50
• Presidential documents narrow by executive_order_number (bare digits, 14192), presidential_document_type (executive_order, proclamation, memorandum, determination, notice, presidential_order, other), president (a slug such as donald-trump), and a signed_after / signed_before window, which differs from the publication window because signing precedes publication. These filters beside a type that leaves out PRESDOCU, or significant beside them or beside type of PRESDOCU alone, fail with type_conflict
• citation ("89 FR 49102") with citation_date (the date its source note prints beside it) returns the documents printed on that page, in page order, volume 59 (1994) onward, narrowed by any other filter. A cite missing its date, or combined with the publication date window, fails with citation_incomplete; a volume before 59 or a date from another year fails with citation_out_of_range. When no document spans the page, the notice gives that day's page range
• Each result carries documentNumber, citation / startPage / endPage, agencies[].slug, the printed docketIds, regulationsGovDocketId and regulationsGovDocumentId, commentCount, commentUrl, commentsCloseOn with commentPeriodOpen (open through 11:59 PM Eastern on the close date), regulationIdNumbers, significant (null when no determination is recorded), presidential (type, number, signingDate, and president; null unless the document is a presidential document), and cfrReferences for the follow-up tools. totalPages and nextPage page the results. The Federal Register serves 50 pages, so truncated marks a set larger than 50 × per_page (1,000 at the default, 5,000 at 100); narrow the date window rather than paging deeper. totalCount stops at 10,000, which means at least that many. A window whose start falls after its end (publication, signing, or effective) fails with date_range_inverted, and invalid_filter names a rejected parameter

regulations_get_document _tool

• document_number in any form the Federal Register issues: 2024-07773 from 2010 on, and older and correction numbers such as 98-1572, E9-25990, or C1-2009-30484; include_full_text adds the plain-text body as one window of max_chars (default 64,000; above 200,000 serves 200,000 with a notice) starting at offset, and passing either of those implies it
• Returns metadata plus the handles other tools take: the Regulations.gov docketId and regulationsGovDocumentId, the printed docketIds, citation / startPage / endPage, commentCount, commentUrl, commentPeriodOpen, significant, presidential for a presidential document, and cfrReferences. With text, fullTextLength and fullTextNextOffset (present while text remains) page the body; a major final rule runs past a million characters

regulations_browse_cfr _tool

• mode: "structure" lists the 50 titles, a title's top-level divisions, or, with title + part, every section and appendix in the part with its subpart and subjectGroup; mode: "search" full-text-searches the codified CFR and requires query; mode: "versions" lists the dates a part's or one section's text changed. title, part (needs title), and a point-in-time date scope structure and search, and per_page from 1 (default 20; above 50 serves 50 with a notice) pages search results, part listings, and versions
• Versions mode takes title with part (and optionally section, written as regulations_get_cfr_section accepts it) or with changed_since (YYYY-MM-DD: versions published by an eCFR issue on or after it, across the whole title when no part is given); any other scope fails with versions_scope_required, and a section citing another title fails with conflicting_title. versions[] rows carry cfrCite, heading, part, subpart, date, issueDate, substantive, and removed, in eCFR's order (by identifier, oldest first within each), and latestAmendmentDate covers the whole filtered set. Pass a row's date as date to regulations_get_cfr_section to read that version: a row dated before 2017-01-01 reads at 2017-01-01, and a removed row's date reads nothing
• Search rows carry cfrCite, heading, hierarchyPath, and excerpt, one per section. source (mirror / live) and sourceScope name the corpus that answered and what it covers, and countBasis says whether totalCount counts sections or section_versions; live search reaches eCFR's first 10,000 hits only (page_out_of_window past them)

regulations_get_cfr_section _tool

• One section (title + part + section), a whole part (section omitted), or one appendix (appendix, verbatim as regulations_browse_cfr emits it, e.g. Appendix A-1 to Part 50). section also takes cites as people write them ("61", "§ 141.61", "141.61(c)") and the cfrCite the server returns ("40 CFR 141.61", "14 CFR 241 § 25"; a cite naming another title fails with conflicting_title), part takes "Part 141", and date reads text from 2017-01-01 through the title's up-to-date date
• Text comes back as one bodyText window (max_chars default 64,000, above 200,000 serves 200,000 with a notice, from offset), paged by bodyTextLength and bodyTextNextOffset. A whole-part read adds a sections[] index with each section's offset and names its appendices without their text; source (mirror / live) reports provenance
• hierarchyPath runs from the title down to the location read. If eCFR cannot supply it, the read still answers, with only the cited levels (Title 40 › Part 141 › § 141.61), and notice says so
• A whole-part read also returns the part's heading, authority, sourceNote, and notes, and each sections[] entry carries the Authority or Source its subpart or subject group states for it
• Superscripts read ^x and subscripts _x (3 × 10^−8, CO_{2}e), footnote markers [n], and diacritics and overlines are combining marks (x̄)

regulations_get_docket _tool · key required

• Exactly one of docket_id (list mode) or document_id (detail mode): neither fails with target_required, both with multiple_targets, and search_term or document_types beside document_id with filter_requires_list_mode. IDs are case-insensitive
• List mode takes docket_id, the Regulations.gov docket ID (e.g. EPA-HQ-OAR-2025-0194) — regulationsGovDocketId on a search or open-comments row, docketId from regulations_get_document, not a printed docketIds entry. search_term keeps the documents whose text matches, ranked by relevance instead of newest first; document_types filters to Proposed Rule, Rule, Notice, Supporting & Related Material, or Other, and combines with it; per_page from 5 (default 25; above 250 serves 250 with a notice), page 1–40
• List mode returns the docket's metadata, documentCount, and documents[], each with an objectId for regulations_find_comments, a frDocNum back to regulations_get_document, and commentPeriodOpen as Regulations.gov reports it. totalPages and nextPage page the documents; Regulations.gov serves 40 pages, so truncated marks a docket larger than 40 × per_page (10,000 at 250)
• Detail mode takes document_id (documentId from a listing, or regulationsGovDocumentId from regulations_get_document) and returns one document with fileFormats (the main file's download URLs), attachments in upload order with withheld ones kept, pageCount, abstract, and restrictReasonType. A document whose content is only in attachments, or that has no file at all, gets a notice saying so

regulations_find_comments _tool · key required

• Exactly one of docket_id, document_object_id, fr_document_number, or comment_id; list scopes take per_page from 5 (default 25; above 250 serves 250 with a notice) and page 1–40
• document_object_id takes an object ID (0900006485883ec6) or a document ID (EPA-HQ-OW-2022-0114-0027). fr_document_number resolves to the Regulations.gov document carrying that exact number and answers not_found when none does. Docket, document, and comment IDs are case-insensitive
• A docket_id that lists no comments is looked up: a docket Regulations.gov does not hold (a printed docket number such as REG-101355-26) answers not_found, and an existing docket answers an empty list with a notice
• Lists narrow by comment text with search_term, and each hit then carries highlightedContent, the matched passages. posted_after / posted_before set an inclusive posted-date window. A backwards window fails with date_range_inverted, and filters passed with comment_id fail with filter_requires_list_mode
• mode is list or detail. Lists return comment summaries without body text. comment_id returns the body, submitter, receivedDate, postmarkDate, duplicateComments (above 1 marks a mass-mail campaign record), and attachments in upload order, with attachmentOnly: true when the substance is in a PDF/DOCX file. An attachment whose file is withheld stays in the list with empty formats and its restrictReasonType (e.g. Copyrighted). Lists carry totalPages and nextPage; truncated marks a set larger than the 40 pages Regulations.gov serves reach (40 × per_page, 10,000 at 250), and posted-date windows take it one slice at a time

regulations_list_open_comments _tool

• Optional query, type (PRORULE, RULE, NOTICE; default ["PRORULE", "RULE"]), agencies (Federal Register slugs), and closing_before; per_page from 1 (default 20; above 100 serves 100 with a notice)
• Rows sort by closing date, soonest first, and carry commentsCloseOn, daysRemaining, documentNumber, the printed docketIds, regulationsGovDocketId and regulationsGovDocumentId, commentCount, and commentUrl. asOf is today's date in Eastern time, so a document stays listed through its close day. totalPages and nextPage page the window; truncated means the Federal Register's 10,000-document limit was reached
• Keyless. keyed reports whether REGULATIONS_GOV_API_KEY is set, which regulations_get_docket and regulations_find_comments need to follow up on a row

regulations://document/{documentNumber} _resource

• documentNumber in any Federal Register form (2024-07773, 98-1572, E9-25990); the payload is regulations_get_document's without the body text
• docketId, regulationsGovDocumentId, the printed docketIds, cfrReferences, citation, commentCount, and commentUrl chain into the search, comment, and CFR tools

regulations://cfr/{title}/{part}/{section} _resource

• Sections only, at the current date and the tool's default 64,000-character window; the part and section segments resolve the way the tool resolves them ("Part 141"; "61", "§ 141.61", "40 CFR 141.61")
• source (mirror / live) reports provenance, and notice says how a part or section written another way resolved and when hierarchyPath holds only the cited levels. bodyTextNextOffset marks a section longer than the window; read the rest through regulations_get_cfr_section with offset

Features

Built on @cyanheads/mcp-ts-core: stdio and Streamable HTTP transports, pluggable auth (none / jwt / oauth), swappable storage (in-memory, filesystem, Supabase, Cloudflare KV/R2/D1), structured logging with optional OpenTelemetry tracing.

Federal Register / eCFR / Regulations.gov-specific:

• One workflow over three official sources: the agent calls regulatory verbs (search_rules, get_cfr_section, find_comments) rather than three API clients
• Cross-source stitching: every Federal Register document surfaces its docket ID and CFR-part handles, which feed the proposal → comments → final rule → codified text trace
• Keyless core: the five Federal Register and eCFR tools need no key. regulations_get_docket and regulations_find_comments need REGULATIONS_GOV_API_KEY (free at api.data.gov/signup, 1,000 requests/hour) and fail with auth_required, naming the variable and signup URL, when it is missing or rejected
• Locally mirrored codified CFR: the eCFR syncs into embedded SQLite + FTS5 for exact-cite reads and full-text search, and falls back to the live API for historical dates, whole parts, appendices, titles outside the mirror, and titles eCFR has re-issued since the mirror last synced them
• A 45-second budget per request, shared by every upstream call, retry, and backoff, so a stalled source answers with upstream_unavailable inside a client's timeout

Agent-friendly output:

• Provenance: source: "mirror" | "live" on every CFR read, plus a sourceScope line on search naming what the corpus covers
• Honest paging: totalPages and nextPage say how far a list goes, a page past the end names the last page instead of reading as "nothing matched", and matches beyond the Federal Register's 50 pages or Regulations.gov's 40 surface through truncated and notice, never as silently dropped rows
• Attachment-aware comments: attachmentOnly flags a comment whose substance is a file, with the download URLs

Getting started

Connect to the hosted instance:

claude mcp add --transport http federal-regulations-mcp-server https://federal-regulations.caseyjhand.com/mcp

Or run it locally:

claude mcp add federal-regulations-mcp-server -- npx -y @cyanheads/federal-regulations-mcp-server

For other clients, add this to your MCP configuration file:

{  "mcpServers": {    "federal-regulations-mcp-server": {      "command": "npx",      "args": ["-y", "@cyanheads/federal-regulations-mcp-server"]    }  }}

Set optional REGULATIONS_GOV_API_KEY to enable docket and comment tools; get a free key at api.data.gov/signup. The Federal Register and eCFR tools work without it. For Docker or a source build, see Running the server.

Configuration

See .env.example for the full list of optional overrides.

Running the server

Local development

Requires Bun v1.4.0 or higher (or Node.js v24+).

git clone https://github.com/cyanheads/federal-regulations-mcp-server.gitcd federal-regulations-mcp-server && bun installcp .env.example .env

• Build and run:

  sh

  # One-time buildbun run rebuild
  # Run the built serverbun run start:stdio# orbun run start:http

• Build or refresh the eCFR mirror (out of band and idempotent; the CFR tools use the live eCFR API until the first build completes):

  sh

  bun run mirror:init      # full build, resumable (scope with ECFR_MIRROR_TITLES)bun run mirror:refresh   # re-harvest every configured title against the latest eCFR issuesbun run mirror:verify    # report row counts and the last-synced issue date

• Run checks and tests:

  sh

  bun run devcheck   # Lint, format, typecheck, security, changelog syncbun run test       # Vitest test suite

Docker

docker run --rm -e REGULATIONS_GOV_API_KEY=your-key -p 3010:3010 ghcr.io/cyanheads/federal-regulations-mcp-server:latest

The Dockerfile defaults to HTTP transport, stateless session mode, and logs to /var/log/federal-regulations-mcp-server. Build the mirror inside a running container with docker exec <container> bun run mirror:init, and mount a volume over /usr/src/app/data so it survives recreation. OpenTelemetry peer dependencies are installed by default; build with --build-arg OTEL_ENABLED=false to omit them.

Project structure

Development guide

See CLAUDE.md for development guidelines and architectural rules. The short version:

• Handlers throw, framework catches — no try/catch in tool logic
• Use ctx.log for request-scoped logging, ctx.state for tenant-scoped storage
• Register new tools and resources via the barrels in src/mcp-server/*/definitions/index.ts
• Wrap external API calls: validate raw → normalize to the domain type → return the output schema; never fabricate missing upstream fields

Data disclaimer

eCFR content is "authoritative but unofficial" per the Office of the Federal Register, not the official legal edition of the Code of Federal Regulations; verify against govinfo.gov for legal research. This server is not affiliated with or endorsed by the Office of the Federal Register, the Government Publishing Office, or the General Services Administration.

Contributing

Issues are welcome. Run checks and tests before submitting:

bun run devcheckbun run test

License

Apache-2.0 — see LICENSE for details.