--- title: "Federal Register scraper skill for AI agents" description: "How an AI agent gets data from Federal Register: the URLs, endpoints and selectors that worked, with code, and the gotchas. An open-source browser-harness skill, last updated Nov 8, 2026." canonical: "https://browser-use.com/skills/federal-register" last_updated: "2026-11-09" --- # Federal Register scraper skill for AI agents Source: https://browser-use.com/skills/federal-register. One file from github.com/browser-use/browser-harness, MIT licensed. Written for browser-harness: code calls its helpers (http_get, js, goto_url). Sites change; these notes describe what worked when last updated. --- # Federal Register - Documents an Agency Published This Week (official API) Field-tested against www.federalregister.gov on 2026-11-08 (a Sunday) using a Browser Use cloud browser, logged out, plus plain HTTP (Python urllib, curl). About 25 browser API calls, 15 page loads and 15 plain HTTP calls, all serial: every answer was HTTP 200 or a documented 400, no 429, no challenge, no login or consent wall. robots.txt (fetched 2026-11-08) has only a `User-agent: *` group, which disallows `/documents/current`, `/documents/email-a-friend`, `/articles/search`, `/documents/search`, `/public-inspection/search`, `/regulations/search`, `/my/` and `/auth/`; `/api/v1/` is not disallowed and there are no named AI-agent groups. Terms: there is no separate ToS (the old terms-of-service URL is a 404); the API page says "No API keys are needed; all you need is an HTTP client or browser", and its one usage restriction is "Republishers of Federal Register material are not permitted to use official NARA or OFR logos or seals." The site's access page (unblock.federalregister.gov) says "programmatic access to these sites is limited to access to our extensive developer APIs", so both blocks read data only from `/api/v1/`. About This Site: "Any person may reproduce or republish any material appearing in any regular or special edition of the Federal Register (1 CFR 2.6)." ## Quick summary - **Use the API, never the HTML search.** `GET /api/v1/documents.json?conditions[agencies][]=&conditions[publication_date][gte]=&conditions[publication_date][lte]=&order=newest` lists the week's documents as JSON. `/documents/search` is disallowed in robots.txt. - **Agency slugs** come from `GET /api/v1/agencies` (473 agencies on 2026-11-08: `slug`, `name`, `short_name`, `parent_id`, `child_slugs`). Names read "Agriculture Department", not "Department of Agriculture". A wrong slug is HTTP 400 `{"errors":{"agencies":"invalid value"}}`, not an empty list. - **Set an end date.** On Sunday 2026-11-08 the API already listed Monday 2026-11-09's issue: EPA `gte=2026-11-02` alone gave 18 documents, with `lte=2026-11-08` 16. - **A parent department includes its sub-agencies.** Each document lists every agency on it (`agriculture-department` plus `animal-and-plant-health-inspection-service`), so both slugs match it. - **Pick fields** with repeated `fields[]=` (56 names in the spec's `DocumentField` enum): `comments_close_on`, `effective_on`, `docket_ids`, `regulation_id_numbers`, `action`, `html_url`, `pdf_url` and more. - **Paging:** `per_page` (default 20; spec says 1000 max), follow `next_page_url` (it carries a `search_after_cursor`). The docs: "you can only paginate through the first 2000 results. If you wish to download more results, use the date range filters." One agency-week is far below that (EPA 16, USDA 18, all agencies 466 for 2026-11-02..08). - **Browser or plain HTTP, API only.** Block 1 opens the tab on an API URL, so `fetch()` is same-origin and no HTML page is loaded. The API also answers plain HTTP (block 2). HTML pages over plain HTTP are redirected to `unblock.federalregister.gov` ("Request Access", a CAPTCHA): the site limits programmatic access to the API. Do not script HTML pages. ## What works ### 1. One agency's documents this week, by name or short name (browser, official API) Self-contained. Resolves `AGENCY` against the agency list, takes this week's Monday to Sunday in Eastern time, and follows `next_page_url` to the end. `fetch_json()` returns parsed JSON on 200, waits 30 s on 429/5xx and otherwise raises, so a failure never looks like "no documents". ```python import collections, datetime, json, os, tempfile, time, urllib.parse, zoneinfo AGENCY = "EPA" # agency name, short name or slug, e.g. "Environmental Protection Agency", "EPA", "environmental-protection-agency" SITE = "https://www.federalregister.gov" API = SITE + "/api/v1/" FIELDS = ["document_number", "publication_date", "type", "title", "action", "agencies", "html_url", "pdf_url", "comments_close_on", "effective_on", "docket_ids", "regulation_id_numbers", "significant"] def on_site(): """A logged-out tab on www.federalregister.gov, so fetch() to /api/v1/ is same-origin. It opens a small API URL, not an HTML page: the site limits programmatic access to its API.""" try: if js("location.origin") == SITE: return except Exception: pass new_tab(API + "documents.json?per_page=1&fields[]=document_number") wait_for_load() time.sleep(2) if js("location.origin") != SITE: raise RuntimeError(f"tab landed on {js('location.href')} (access check?): stop and tell the user; do not retry around it") # js() gives up after a few seconds, so start the fetch, then poll for it. START_JS = """(() => { window.__fr = window.__fr || {}; const id = String(Math.random()).slice(2); window.__fr[id] = null; fetch(%s, {credentials: 'omit'}) .then(async r => { window.__fr[id] = {status: r.status, url: r.url, type: r.headers.get('content-type'), text: await r.text()}; }) .catch(e => { window.__fr[id] = {status: 0, url: '', type: '', text: String(e)}; }); return id; })()""" POLL_JS = """(() => { const v = window.__fr[%s]; if (v) delete window.__fr[%s]; return v; })()""" def fetch_json(url, tries=3, timeout=60): """GET url from inside the tab -> parsed JSON. 400 carries {"errors": {...}}. Waits 30 s on 429/5xx, raises on anything else, and on an HTML answer (the site's "Request Access" check), so a failure never looks like "no documents".""" for attempt in range(tries): rid = json.dumps(js(START_JS % json.dumps(url))) t0 = time.time() while (r := js(POLL_JS % (rid, rid))) is None: if time.time() - t0 > timeout: raise RuntimeError(f"{url}: no answer after {timeout}s") time.sleep(0.5) time.sleep(1) # serial, about one call per second if r["status"] == 200 and "json" in (r["type"] or ""): return json.loads(r["text"]) if r["status"] in (0, 429, 500, 502, 503, 504) and attempt < tries - 1: time.sleep(30) continue raise RuntimeError(f"{url}: HTTP {r['status']} {r['type']} at {r['url']}: {r['text'][:200]!r}; stopping") def find_agency(q, agencies): """Exact slug or name, then short name, then a unique substring of the name. Raises with candidates otherwise: short names repeat (DOE = Energy Department and its EERE office, FS = Fiscal Service and Forest Service).""" ql = q.strip().lower() for a in agencies: if ql in (a["slug"], a["name"].lower()): return a hits = [a for a in agencies if (a.get("short_name") or "").lower() == ql] or [a for a in agencies if ql in a["name"].lower()] if len(hits) == 1: return hits[0] raise ValueError(f"{q!r}: {len(hits)} matches: {[(a['slug'], a['name']) for a in hits[:10]]}") def documents(slug, since, until=None): """Every document the agency (or any of its sub-agencies) published on or after `since`, newest first, following next_page_url.""" params = [("conditions[agencies][]", slug), ("conditions[publication_date][gte]", since), ("order", "newest"), ("per_page", 100)] if until: params.append(("conditions[publication_date][lte]", until)) params += [("fields[]", f) for f in FIELDS] url = API + "documents.json?" + urllib.parse.urlencode(params) rows, first = [], None while url: d = fetch_json(url) first = first or d if d["count"] > 2000: raise RuntimeError(f"{d['count']} documents: the API pages through the first 2000 only, and its docs say to use the date range filters for more") rows += d.get("results", []) # a 0-count answer has no "results" key url = d.get("next_page_url") if len(rows) != first["count"]: raise RuntimeError(f"got {len(rows)} rows, count says {first['count']}: paging changed; stopping") return first["description"], rows on_site() agencies = fetch_json(API + "agencies") ag = find_agency(AGENCY, agencies) today = datetime.datetime.now(zoneinfo.ZoneInfo("America/New_York")).date() # the Federal Register publishes on Eastern time monday = today - datetime.timedelta(days=today.weekday()) sunday = monday + datetime.timedelta(days=6) # without an end date the next issue shows up too: it is listed before its publication date desc, docs = documents(ag["slug"], monday.isoformat(), sunday.isoformat()) print(f"{len(agencies)} agencies listed; {AGENCY!r} -> {ag['slug']} ({ag['name']}, short name {ag.get('short_name')}, {len(ag['child_slugs'])} sub-agencies)") print(f"today {today} (ET), week {monday}..{sunday}: {desc}") print(f"{len(docs)} documents | by type {dict(collections.Counter(d['type'] for d in docs))}") print("by day:", dict(collections.Counter(d["publication_date"] for d in docs))) for d in docs: extra = " | ".join(x for x in (d["comments_close_on"] and f"comments close {d['comments_close_on']}", d["effective_on"] and f"effective {d['effective_on']}", d["significant"] and "significant") if x) print(f"\n{d['publication_date']} {d['type']:<13} {d['document_number']} {d['title'][:100].rstrip()}") print(f" {(d['action'] or '-')[:60]} | dockets {d['docket_ids'][:2]} | RIN {d['regulation_id_numbers'][:1]}" + (f" | {extra}" if extra else "")) print(" ", d["html_url"]) fd, out = tempfile.mkstemp(prefix=f"federal_register_{ag['slug']}_{monday}_", suffix=".json") # a new file each run with os.fdopen(fd, "w") as f: json.dump(docs, f) print("\nsaved:", os.path.basename(out), "in tempfile.gettempdir()") ``` Output (from the run on 2026-11-08): ```text 473 agencies listed; 'EPA' -> environmental-protection-agency (Environmental Protection Agency, short name EPA, 0 sub-agencies) today 2026-11-08 (ET), week 2026-11-02..2026-11-08: Documents published from 11/02/2026 to 11/08/2026 and from Environmental Protection Agency 16 documents | by type {'Notice': 6, 'Rule': 5, 'Proposed Rule': 5} by day: {'2026-11-06': 2, '2026-11-05': 10, '2026-11-03': 2, '2026-11-02': 2} 2026-11-06 Notice 2026-22684 Environmental Impact Statements; Notice of Availability - | dockets ['FRL OPRM-FAD-247'] | RIN [] https://www.federalregister.gov/documents/2026/11/06/2026-22684/environmental-impact-statements-notice-of-availability 2026-11-06 Rule 2026-22671 Air Plan Approval; California; Revised Format for Materials Incorporated by Reference Final rule; administrative change. | dockets ['EPA-R09-OAR-2024-0527', 'FRL-13582-01-R9'] | RIN [] | effective 2026-11-06 https://www.federalregister.gov/documents/2026/11/06/2026-22671/air-plan-approval-california-revised-format-for-materials-incorporated-by-reference 2026-11-05 Rule 2026-22647 Air Plan Revisions; Arizona; Maricopa County Air Quality Department; New Source Review; Emission Red Final rule. | dockets ['EPA-R09-OAR-2026-1684', 'FRL-13268-02-R9'] | RIN [] | effective 2026-12-07 https://www.federalregister.gov/documents/2026/11/05/2026-22647/air-plan-revisions-arizona-maricopa-county-air-quality-department-new-source-review-emission 2026-11-05 Proposed Rule 2026-22592 Air Plan Approval; District of Columbia; Enhanced Vehicle Emission Inspection and Maintenance (I/M) Proposed rule. | dockets ['EPA-R03-OAR-2026-1024', 'FRL-12072-01-R3'] | RIN [] | comments close 2026-12-07 https://www.federalregister.gov/documents/2026/11/05/2026-22592/air-plan-approval-district-of-columbia-enhanced-vehicle-emission-inspection-and-maintenance-im 2026-11-05 Proposed Rule 2026-22588 Air Quality Implementation Plans; California; San Joaquin Valley; 2015 8-Hour Ozone Nonattainment Ar Proposed rule. | dockets ['EPA-R09-OAR-2026-8252', 'FRL-13635-01-R9'] | RIN [] | comments close 2026-12-07 https://www.federalregister.gov/documents/2026/11/05/2026-22588/air-quality-implementation-plans-california-san-joaquin-valley-2015-8-hour-ozone-nonattainment-area 2026-11-05 Proposed Rule 2026-22585 Approval and Promulgation of State Plans for Designated Facilities and Pollutants; Arizona; Control Proposed rule. | dockets ['EPA-R09-OAR-2026-5710', 'FRL-13516-01-R9'] | RIN [] | comments close 2026-12-07 https://www.federalregister.gov/documents/2026/11/05/2026-22585/approval-and-promulgation-of-state-plans-for-designated-facilities-and-pollutants-arizona-control-of 2026-11-05 Rule 2026-22584 Approval and Promulgation of State Plans for Designated Facilities and Pollutants; West Virginia; Co Final rule. | dockets ['EPA-R03-OAR-2023-0230', 'FRL-9329-03-R3'] | RIN [] | effective 2026-12-07 https://www.federalregister.gov/documents/2026/11/05/2026-22584/approval-and-promulgation-of-state-plans-for-designated-facilities-and-pollutants-west-virginia 2026-11-05 Proposed Rule 2026-22583 Approval and Promulgation of Delegation of Authority for Designated Facilities and Pollutants; Alleg Proposed rule. | dockets ['EPA-R03-OAR-2026-5021', 'FRL-13501-01-R3'] | RIN [] | comments close 2026-12-07 https://www.federalregister.gov/documents/2026/11/05/2026-22583/approval-and-promulgation-of-delegation-of-authority-for-designated-facilities-and-pollutants 2026-11-05 Rule 2026-22576 Significant New Use Rules on Certain Chemical Substances (25-3.5e) Final rule. | dockets ['EPA-HQ-OPPT-2025-2169', 'FRL-13126-03-OCSPP'] | RIN ['2070-AB27'] | effective 2027-01-04 https://www.federalregister.gov/documents/2026/11/05/2026-22576/significant-new-use-rules-on-certain-chemical-substances-25-35e 2026-11-05 Notice 2026-22548 Proposed Consent Decree Notice of proposed consent decree; request for public commen | dockets ['EPA-HQ-OGC-2026-9242', 'FRL-13676-01-OGC'] | RIN [] | comments close 2026-12-07 https://www.federalregister.gov/documents/2026/11/05/2026-22548/proposed-consent-decree 2026-11-05 Notice 2026-22547 Public Meeting of the Clean Air Scientific Advisory Committee (CASAC) on Lead Notice. | dockets ['FRL- 13687-01-AO'] | RIN [] https://www.federalregister.gov/documents/2026/11/05/2026-22547/public-meeting-of-the-clean-air-scientific-advisory-committee-casac-on-lead 2026-11-05 Notice 2026-22546 Cross-Media Electronic Reporting: Announcement of Determination To Approve Authorized Program Revisi Notice. | dockets ['FRL-13494-01-OFA'] | RIN [] https://www.federalregister.gov/documents/2026/11/05/2026-22546/cross-media-electronic-reporting-announcement-of-determination-to-approve-authorized-program 2026-11-03 Rule 2026-22404 National Emission Standards for Hazardous Air Pollutants: Lime Manufacturing Plants Technology Revie Correcting amendment. | dockets ['EPA-HQ-OAR-2026-7525', 'FRL-5948.4-01-OAR'] | RIN [] | effective 2026-11-03 https://www.federalregister.gov/documents/2026/11/03/2026-22404/national-emission-standards-for-hazardous-air-pollutants-lime-manufacturing-plants-technology-review 2026-11-03 Notice 2026-22392 Avanti Corporation; Access to Confidential Business Information Notice. | dockets ['EPA-HQ-OPPT-2003-0004', 'FRL-13681-01-OCSPP'] | RIN [] https://www.federalregister.gov/documents/2026/11/03/2026-22392/avanti-corporation-access-to-confidential-business-information 2026-11-02 Proposed Rule 2026-22295 Significant New Use Rules on Certain Chemical Substances (26-7) Proposed rule. | dockets ['EPA-HQ-OPPT-2026-5974', 'FRL-13610-01-OCSPP'] | RIN ['2070-AB27'] | comments close 2026-12-02 https://www.federalregister.gov/documents/2026/11/02/2026-22295/significant-new-use-rules-on-certain-chemical-substances-26-7 2026-11-02 Notice 2026-22291 Certain New Chemicals or Significant New Uses; Statements of Findings-July 2026 Notice. | dockets ['EPA-HQ-OPPT-2026-1849', 'FRL-13203-07-OCSPP'] | RIN [] https://www.federalregister.gov/documents/2026/11/02/2026-22291/certain-new-chemicals-or-significant-new-uses-statements-of-findings-july-2026 saved: federal_register_environmental-protection-agency_2026-11-02_r6qb0qti.json in tempfile.gettempdir() ``` ### 2. A department with sub-agencies: breakdown, comment deadlines, next issue (plain HTTP, no browser) Python standard library only, truthful User-Agent, 4 calls at 1 s spacing. ```python import collections, datetime, json, time, urllib.error, urllib.parse, urllib.request, zoneinfo AGENCY = "agriculture-department" # a parent department: its sub-agencies' documents are included UA = {"User-Agent": "browser-harness-skill/1.0 (Python urllib)"} API = "https://www.federalregister.gov/api/v1/" def call(path, params=()): """GET -> parsed JSON. 400 = bad parameter ({"errors": {...}}), raised as ValueError. Waits 30 s on 429/5xx, then stops. An HTML answer means the site's access check (unblock.federalregister.gov): stop, do not work around it.""" url = API + path + ("?" + urllib.parse.urlencode(list(params)) if params else "") for attempt in range(3): try: with urllib.request.urlopen(urllib.request.Request(url, headers=UA), timeout=60) as r: status, body, ctype, final = r.status, r.read(), r.headers.get("content-type", ""), r.url except urllib.error.HTTPError as e: status, body, ctype, final = e.code, e.read(), e.headers.get("content-type", ""), url time.sleep(1) # serial, about one call per second if status == 200 and "json" in ctype: return json.loads(body) if status == 400: raise ValueError(f"HTTP 400 {body[:200].decode()}") if status not in (429, 500, 502, 503, 504) or attempt == 2: raise RuntimeError(f"{url}: HTTP {status} {ctype} at {final}; stopping") time.sleep(30) def week_docs(slug, since, until, fields): params = [("conditions[agencies][]", slug), ("conditions[publication_date][gte]", since), ("conditions[publication_date][lte]", until), ("order", "newest"), ("per_page", 1000)] + [("fields[]", f) for f in fields] d = call("documents.json", params) if d["count"] > 1000: raise RuntimeError(f"{d['count']} documents: page with next_page_url or split the dates") return d.get("results", []) ag = call("agencies/" + AGENCY) today = datetime.datetime.now(zoneinfo.ZoneInfo("America/New_York")).date() monday = today - datetime.timedelta(days=today.weekday()) docs = week_docs(AGENCY, monday.isoformat(), (monday + datetime.timedelta(days=6)).isoformat(), ["document_number", "publication_date", "type", "title", "agencies", "comments_close_on", "html_url"]) print(f"{ag['name']}: {len(ag['child_slugs'])} sub-agencies | week {monday}..{monday + datetime.timedelta(days=6)}: {len(docs)} documents") by_sub = collections.Counter(" + ".join(a["name"] for a in d["agencies"] if a.get("slug") != AGENCY) or ag["name"] + " (itself)" for d in docs) # a document can list several sub-agencies print("by sub-agency:", by_sub.most_common()) print("by type:", dict(collections.Counter(d["type"] for d in docs))) open_comments = sorted((d["comments_close_on"], d["document_number"], d["title"]) for d in docs if d["comments_close_on"]) print(f"\n{len(open_comments)} with a comment deadline:") for close, num, title in open_comments: print(f" closes {close} {num} {title[:90].rstrip()}") # The next issue is already listed before its date: documents dated after today nxt = week_docs(AGENCY, (today + datetime.timedelta(days=1)).isoformat(), (today + datetime.timedelta(days=7)).isoformat(), ["document_number", "publication_date", "type", "title"]) print(f"\nalready listed with a later date: {len(nxt)}", [(d["publication_date"], d["type"], d["title"][:60]) for d in nxt[:3]]) # A wrong slug is a 400, not an empty list try: call("documents.json", [("conditions[agencies][]", "usda"), ("conditions[publication_date][gte]", monday.isoformat())]) except ValueError as e: print("\nslug 'usda':", e) ``` Output (from the run on 2026-11-08): ```text Agriculture Department: 36 sub-agencies | week 2026-11-02..2026-11-08: 18 documents by sub-agency: [('Agriculture Department (itself)', 6), ('Food and Nutrition Administration', 5), ('Foreign Agricultural Service', 3), ('Agricultural Marketing Service', 2), ('Animal and Plant Health Inspection Service', 1), ('Forest Service', 1)] by type: {'Notice': 16, 'Rule': 2} 6 with a comment deadline: closes 2027-01-04 2026-22309 Notice of Request for Revision to and Extension of Approval of an Information Collection; closes 2027-01-04 2026-22438 Notice of Request for an Extension of a Currently Approved Information Collection closes 2027-01-04 2026-22442 Notice of Request for an Extension of a Currently Approved Information Collection closes 2027-01-04 2026-22445 Notice of Request for Approval of an Existing Collection for Borlaug Fellowship Program OM closes 2027-01-04 2026-22467 Assessment Exemption for Organic Commodities Under Federal Marketing Orders; Notice of Req closes 2027-01-05 2026-22679 Special Supplemental Nutrition Program for Women, Infants, and Children (WIC): Online Orde already listed with a later date: 3 [('2026-11-09', 'Proposed Rule', 'Pecans Grown in the States of Alabama, Arkansas, Arizona, Ca'), ('2026-11-09', 'Notice', 'Pioneer Hi-Bred International, Inc.: Availability of a Petit'), ('2026-11-09', 'Rule', 'Single Family Housing Guaranteed Loan Program-Limited Party ')] slug 'usda': HTTP 400 {"errors":{"agencies":"invalid value"}} ``` ## URL patterns (all observed) | URL | What | |---|---| | `/api/v1/documents.json?conditions[agencies][]=&conditions[publication_date][gte]=YYYY-MM-DD&conditions[publication_date][lte]=YYYY-MM-DD&order=newest&per_page=N&fields[]=...` | the week's documents; repeat `conditions[agencies][]` for several agencies (OR: with `gte=2026-11-02` only, EPA 18 + APHIS 2 = 20) | | `/api/v1/documents/facets/type?conditions[...]` | counts by type: `{"NOTICE":{"count":6,"name":"Notice"},"PRORULE":{...},"RULE":{...}}` (EPA, 2026-11-02..08) | | `/api/v1/agencies`, `/api/v1/agencies/` | agency list / one agency (`child_slugs`, `parent_id`, `url`) | | `/api/v1/documentation.json` | the OpenAPI spec behind the interactive docs page (parameter and field enums, agency slug enum) | | `/documents/YYYY/MM/DD//` | `html_url`, the document page for people: link to it, do not script it (read the document's fields from the API) | ## Fields | Field | Notes | |---|---| | `document_number` | `2026-22788`; the stable ID | | `publication_date` | `YYYY-MM-DD`; issues are weekdays only (EPA had none on 2026-11-04) | | `type` | `Rule`, `Proposed Rule`, `Notice` seen; filter codes `RULE`, `PRORULE`, `NOTICE`, `PRESDOCU` in `conditions[type][]` | | `action` | free text, e.g. `Final rule.`, `Correcting amendment.`; can be `null` | | `agencies[]` | `{name, slug, id, url, json_url, parent_id, raw_name}`; parent and sub-agency both listed | | `comments_close_on`, `effective_on` | `YYYY-MM-DD` or `null` | | `docket_ids[]`, `regulation_id_numbers[]` | `EPA-HQ-OAR-2002-0037`, RIN `2060-AR73`; docket strings are as printed (`FRL- 13687-01-AO`) | | `html_url`, `pdf_url` | document page; PDF on www.govinfo.gov (`/content/pkg/FR-2026-11-02/pdf/2026-22309.pdf`) | | `count`, `total_pages`, `next_page_url`, `description` | top level; `results` is absent when `count` is 0 | ## Gotchas - **Next issue leaks into "this week".** Documents for the next publication date are listed before that date (Monday 2026-11-09 documents on Sunday 2026-11-08). Always pass `conditions[publication_date][lte]`. - **Short names repeat.** `DOE` = Energy Department and Energy Efficiency and Renewable Energy Office, `DOL`, `FS` (Fiscal Service, Forest Service): 10 repeated short names; 53 of 473 agencies have no short name. Block 1 refuses an ambiguous match. - **`count` tops out at 10000.** `gte=2026-01-01` said 10000 while Oct 1 - Nov 8 alone was 2577. Narrow dates before trusting a count. - **Loose date parsing:** `gte=11/02/2026` returned the same 18 as `2026-11-02`. Send ISO dates anyway. - **Tab marker.** The harness prefixes the tab title with a horse emoji; block 1 reads only the origin. ## What does not work - **`/documents/search` (and `/articles/search`, `/documents/current`)**: disallowed in robots.txt. Not used. - **HTML pages for data**: over plain HTTP they redirect to `unblock.federalregister.gov` "Request Access" (a CAPTCHA), which says programmatic access is limited to the API. Do not solve it or work around it; use the API. - **Not tested:** public inspection endpoints, `per_page` above 1000 with more than 1000 results, rate limits (none documented, none hit). --- BROWSER USE - Browser Use Agents: give Browser Use a task and receive completed work. API V4 is current for new integrations. - Browser Infrastructure: connect your agent or automation to managed browsers through SDK, REST, or CDP. Starts at $0.02/browser-hour. - Developer tools: Open Source, Browser Harness, SDK, and MCP support the two products above. [Developer Index](https://browser-use.com/index.md) [Product Map](https://browser-use.com/llms.txt) [Full Product Context](https://browser-use.com/llms-full.txt) [Pricing](https://browser-use.com/pricing.md) [Cloud Docs](https://docs.browser-use.com/cloud/quickstart) [Open Source Docs](https://docs.browser-use.com/open-source/introduction)