USPTO Patent Scraper

Search USPTO patents and published applications by keyword, inventor, assignee or field-coded query — without a login or API key — and get structured patent records with dates, parties, classifications and family IDs, plus optional abstract, claims and description text.

Built for:

• Patent landscape and prior-art research
• Competitor and technology monitoring by assignee or keyword
• Building patent databases for analytics, search or ML
• Tracking newly published applications and grants in a technology area
• Enriching R&D and legal workflows with structured patent metadata

Search US Patents & Prior Art Without an API Key

The actor queries three USPTO collections in one run — granted patents (USPAT), published applications (US-PGPUB) and OCR patents (USOCR) — using the same fast JSON interface as the official Patent Public Search web app. Queries accept plain keywords and USPTO field codes (7557042.pn., microsoft.as., black.in., lithium battery AND @pd>=20240101). Each result row carries the patent number and GUID, title, database, kind code, publication/filing/priority dates, application number, assignee, applicants, inventors, CPC/IPC/USPC classifications and the USPTO family ID. With details enabled, the actor also fetches the full document record: abstract, claims and/or description as plain text, examiner and attorney names, government-interest statement, priority country and claim counts. Results page through up to 20,000 documents per query — the Patent Public Search limit.

How It Works

The Actor sends each query to the same internal JSON endpoint used by Patent Public Search, 500 results per request. With details enabled it fetches the full document and selected text sections; request starts are serialized and HTTP 429 responses trigger a global cooldown.

Each row covers one patent document with number, title, database, dates, assignee, inventors, CPC and IPC classes, and optional abstract, claims and description.

1. The search runs in batches of 500 results per request against the same internal JSON endpoint the official Patent Public Search application calls.
2. Each result row is pushed according to your pushMode (incremental, page or end), so you can watch results appear or collect them in bulk.
3. If fetchDetails is enabled, one extra request per patent retrieves the full document record and selected text sections.
4. Request starts are serialized with a configurable minimum delay (default 200 ms), network errors and 5xx responses are retried, and the USPTO session is rotated automatically; on HTTP 429 the actor applies a global cooldown.
5. A 100-patent run takes roughly 30–40 seconds with details enabled; with fetchDetails: false only one request per 500 results is needed.

Input Parameters

Field	Type	Required	Default	Notes
searchQuery	string	Yes	—	USPTO Patent Public Search query. Plain keywords or field suffixes: litium battery, 7557042.pn., microsoft.as., lithium battery AND @pd>=20240101.
maxItems	integer	No	100	Patents written to the dataset (1–20,000; Patent Public Search caps any query at 20,000).
sort	enum	No	newest	newest, oldest or relevance.
databases	array	No	all three	US-PGPUB (published applications), USPAT (granted patents), USOCR (OCR patents).
defaultOperator	enum	No	OR	How words combine when no explicit operator: OR (broader) or AND (all words).
fetchDetails	boolean	No	true	Fetch the full document record per patent — adds inventors, examiners, attorney and selected text sections; one extra request (and one detail charge) per successfully fetched document.
pushMode	enum	No	incremental	incremental (row by row), page (per 500-result page), end (single batch at the end; more memory).
textSections	array	No	["abstract"]	Which text sections to include when details are fetched: abstract, claims, description (description can be very large).
concurrency	integer	No	8	Parallel detail requests (1–20); the minimum request delay is still enforced.
requestDelayMs	integer	No	200	Minimum delay between USPTO request starts, 100–10,000 ms. The default is tuned for reliable detail runs.

Output Fields

One dataset item per patent document — 34 fields:

• Identity: guid, patentNumber, documentId, title, database, documentType, kindCode, applicationNumber, familyId, detailsFetched
• Dates: publicationDate, filingDate, priorityDate, priorityCountryCode*
• Parties: assignee, assigneeCity, assigneeState, assigneeCountry, applicants, inventors, inventorsShort, primaryExaminer, assistantExaminer, attorneyName*
• Classifications: cpcCodes, cpcAdditionalCodes, ipcCodes, uspcCodes
• Text (when the section is selected): abstract, claims, description
• Other: governmentInterest, numberOfClaims, exemplaryClaimNumber*

* Requires fetchDetails: true.

Example Input

{
  "searchQuery": "lithium battery AND @pd>=20240101",
  "maxItems": 100,
  "sort": "newest",
  "databases": ["US-PGPUB", "USPAT"],
  "fetchDetails": true,
  "textSections": ["abstract"],
  "concurrency": 8,
  "requestDelayMs": 200
}

Fast metadata-only variant:

{ "searchQuery": "solid state battery", "maxItems": 1000, "fetchDetails": false, "pushMode": "page" }

Example Output

{
  "guid": "US-20260316516-A1",
  "patentNumber": "20260316516",
  "documentId": "US 20260316516 A1",
  "title": "Solid-State Battery Cell With Ceramic Separator",
  "database": "US-PGPUB",
  "documentType": "Utility Patent Application",
  "kindCode": "A1",
  "publicationDate": "2026-10-15",
  "filingDate": "2025-04-11",
  "priorityDate": "2024-04-12",
  "applicationNumber": "18/634,210",
  "assignee": "Example Energy Corp.",
  "applicants": ["Example Energy Corp."],
  "inventors": ["Jane Q. Example"],
  "cpcCodes": ["H01M 50/403", "H01M 10/0562"],
  "familyId": 81234567,
  "abstract": "A solid-state battery cell includes a ceramic separator ...",
  "numberOfClaims": 18,
  "detailsFetched": true
}

Automate Patent Monitoring With Schedules & Webhooks

• Weekly new-publications feed: schedule a query such as microsoft.as. AND @pd>=20260101 (adjust the date each run) and keep rows whose patentNumber has not appeared before — a lightweight new-patent alert system.
• Competitor tracking: one scheduled run per assignee, with a dataset webhook pushing matches to Slack, email or your warehouse.
• Landscape snapshots: run broad keyword searches monthly with fetchDetails: false and pushMode: "page" for cheap, fast classification data (cpcCodes) to chart technology trends.
• Scheduled runs are the intended pattern and improve this actor's Store recommendation signals.

Related Actors

• Google Patents Scraper — Search, Citations, Family Graph — the most-used patent actor on the Store, useful for citation and family-graph views.
• Google Patents Scraper — Search, Citations, Family Graph — same coverage from a different developer, a solid cross-check.
• Google Patents Scraper — lighter Google Patents alternative.
• USPTO Patent Search Scraper — No Login — direct USPTO-side extraction with a different field set.
• USPTO Patent Scraper — another USPTO-focused option for redundancy.

Frequently Asked Questions

Why use this actor instead of the official API?

USPTO's Open Data Portal (data.uspto.gov) does offer official APIs, but access has become heavyweight: since June 18, 2026 you need a signed-in USPTO.gov account with multi-factor authentication, and API keys for accounts missing the required profile fields were deactivated on October 8, 2026. The PatentsView APIs have been interrupted since their March 20, 2026 migration with no published relaunch date. This actor sidesteps all of that: it uses the same JSON interface as the official Patent Public Search web app, needs no account, key or registration, and returns structured metadata with optional abstract/claims/description — with keyless access being the point.

Is there a USPTO API?

Yes — the Open Data Portal provides patent search and bulk dataset APIs, but they require a USPTO.gov account, MFA and an API key. PatentsView's legacy APIs are in transition (interrupted since March 20, 2026, relaunch unannounced). Third-party PatentsView-compatible gateways also exist (e.g. patentgrep). This actor offers the keyless route through the public search interface the official web app itself uses.

How can I search patents without an API key?

Run this actor with a query — that is all that is required. It talks to the same public endpoint the official Patent Public Search application uses, rotates its session, and paces requests so large detail runs stay under USPTO's HTTP 429 limit.

What are the alternatives to this data source?

• USPTO Open Data Portal — official APIs and bulk datasets, with account/MFA/key requirements.
• Google Patents — broad coverage (actors linked above) including citations and family graphs.
• PatentsView-compatible gateways — third-party services that mirror PatentsView's API shape.
• Espacenet / EPO — international coverage beyond US documents.

Can I get patent abstracts, claims and full descriptions?

Yes — keep fetchDetails: true and pick sections in textSections (abstract, claims, description). The description section can be very large, so it is excluded by default; each successfully fetched detail adds one extra request and, with pay-per-event pricing, one detail charge.

Does it require a USPTO login?

Not currently — the actor uses the anonymous session endpoint of Patent Public Search. USPTO has announced a sign-in requirement for the Patent Public Search web app from November 7, 2026 (per the actor's documentation); if the anonymous endpoint begins returning 401, the actor reports a clear error instead of failing silently.

For AI Agents & LLM Apps

Purpose: returns USPTO patent records (identity, dates, parties, classifications, family ID) with optional abstract/claims/description text for a given search query — keyless.

Minimal input:

{ "searchQuery": "lithium battery", "maxItems": 5 }

Metadata-only (fast/cheap) variant:

{ "searchQuery": "microsoft.as.", "maxItems": 100, "fetchDetails": false }

Output fields: guid, patentNumber, documentId, title, database, documentType, kindCode, publicationDate, filingDate, priorityDate, priorityCountryCode, applicationNumber, assignee, assigneeCity, assigneeState, assigneeCountry, applicants, inventors, inventorsShort, cpcCodes, cpcAdditionalCodes, ipcCodes, uspcCodes, familyId, abstract, claims, description, primaryExaminer, assistantExaminer, attorneyName, governmentInterest, numberOfClaims, exemplaryClaimNumber, detailsFetched.

Behaviors an agent should know:

• searchQuery is the only required input; maxItems defaults to 100 and can never exceed 20,000 (Patent Public Search's per-query ceiling).
• fetchDetails defaults to true — each detail adds one request and a Patent detail charge; set it to false for metadata-only runs at one request per 500 results.
• Detail-only fields (inventors, examiners, attorney, text sections, city/state/country) are populated only when fetchDetails is on; check detailsFetched to confirm.
• pushMode: "end" buffers everything in memory — prefer incremental or page for large runs.
• Rate limiting is built in: serialized request starts with a 200 ms default delay, automatic session rotation and a global cooldown on HTTP 429. Lower requestDelayMs only if you accept 429 risk.

Billing: pay-per-event — Patent at $0.005 per patent result and Patent detail at $0.001 per successfully fetched document, plus a run-start event.

Legal & Compliance Disclaimer

This actor is an independent tool and is not affiliated with, endorsed by, or sponsored by the United States Patent and Trademark Office (USPTO). It queries only USPTO's public Patent Public Search interface — no account credentials are used — and paces requests to stay under USPTO's rate limits. Patent documents and bibliographic data are public records; users are responsible for their own compliance with USPTO's terms of use and applicable law when storing, republishing or automating access to the data. This actor is a data tool, not legal advice.

SEO Keywords

uspto scraper, uspto patent scraper, patent scraper, patent search api, uspto api, patentsview alternative, prior art search, patent data api, google patents alternative, patent database, us patent search, patent full text, patent abstract data, patent classification data, cpc codes, patent landscape analysis, freedom to operate data, patent monitoring, newly published patents, patent research tool