Google Patents Scraper — Search, Citations, Family Graph

Scrape Google Patents — keyword search, publication IDs, assignee and inventor portfolios, family graphs, and citation hops — and get one structured row per publication: title, abstract, dates, claims, CPC, citations, family members, expiry, and a PDF link. Paste a patents.google.com URL or a query. No API key.

How it works

✨ Why use this scraper?

• Six modes. Keyword search, ID lookup, assignee portfolio, inventor portfolio, family graph, and N-hop citation crawl — same surface as the Store leader, plus URL input.
• Paste a URL. Search pages (?q=, ?assignee=, ?inventor=) and detail pages (/patent/US10000000B2/en) are classified automatically.
• Arrays, not singles. Multiple queries, assignees, inventors, or IDs in one run.
• CPC filter. Restrict search with G06N or a full leaf like G01S7/4863.
• Deep records. Claims, forward/backward citations, family ID + members, CPC leaves, full description, expiry, legal events, PDF.
• No key, no browser. Search hits Google's public /xhr/query JSON; details parse server-rendered HTML.

🎯 Use cases

Who	What they use it for
Patent attorneys & agents	Prior-art search, claim review, family coverage across offices.
IP / competitive intel	Assignee and inventor portfolios, citation neighborhoods around a seed.
R&D and product teams	Landscape a technology area (query + CPC + date) before filing.
Legal-tech & AI agents	Structured patent records for MCP / RAG / landscaping tools.

📥 Supported inputs

You paste	What happens
https://patents.google.com/?q=lidar	Keyword search
https://patents.google.com/?assignee=Apple+Inc.	Assignee portfolio
https://patents.google.com/?inventor=Geoffrey+Hinton	Inventor portfolio
https://patents.google.com/patent/US10000000B2/en	Detail / family / citation seed
US10000000B2 / US-10000000-B2	Same as a detail URL

Not supported: Google Scholar-only literature, login-gated office dockets, or uploading a PDF for OCR.

🔄 How it works

1. Choose a mode and give it queries, IDs, names, or start URLs.
2. Search pages the public xhr/query JSON (100 hits per page).
3. Details fetch /patent/{id}/{lang} and read itemprop fields (claims, family, citations, CPC, expiry).
4. Family / citations walk those fields and, when depth is deep, fetch each member or hop.
5. Export one row per publication as JSON, CSV, or Excel.

⚙️ Input parameters

Field	Type	Description
mode	enum	search, details, byAssignee, byInventor, family, citationNetwork.
startUrls	array	Optional patents.google.com URLs. Classified automatically.
searchQueries / searchQuery	array / string	Free-text queries for search. Boolean AND/OR, quotes, wildcards.
patentIds	array	Publication numbers for details, family, citationNetwork.
assigneeNames / assigneeName	array / string	Company names for byAssignee.
inventorNames / inventorName	array / string	Person names for byInventor.
countryCodes	array	Office codes (US, EP, WO, JP…).
cpc	array	CPC prefixes.
dateFrom / dateTo	string	YYYY-MM-DD bounds.
status	enum	ANY, GRANT, APPLICATION.
patentType	enum	ANY, PATENT, DESIGN, OTHER.
language	string	Detail language (en, de, ja, zh). Default en.
maxItems	integer	Hard cap. Default 50. Caps citation nodes too.
enrichmentDepth	enum	basic or deep.
citationDirection	enum	backward, forward, both.
citationDepth	integer	Hops, 1–3.
proxy	object	Default Apify residential US. Direct works from many networks.

Example input

{
  "mode": "search",
  "searchQueries": ["\"large language model\" training"],
  "countryCodes": ["US", "EP"],
  "dateFrom": "2023-01-01",
  "status": "GRANT",
  "maxItems": 25,
  "enrichmentDepth": "basic"
}

📊 Output overview

One dataset row per publication. Basic search rows carry id, title, snippet, dates, assignee, inventors, and the Google Patents URL. Deep (or any details/family/citation run) adds abstract, description, claims, CPC, citations, family, expiry, legal events, and PDF.

📦 Output sample

{
  "patentId": "US10000000B2",
  "title": "Coherent LADAR using intra-pixel quadrature detection",
  "assignee": "Raytheon Co",
  "inventors": ["Joseph Marron"],
  "filingDate": "2015-03-10",
  "publicationDate": "2018-06-19",
  "expirationDate": "2036-05-13",
  "countryCode": "US",
  "kindCode": "B2",
  "claimsCount": 20,
  "familyId": "55456961",
  "familyMembers": ["US10000000B2", "US10845468B2", "EP3268771B1", "WO2016144528A1"],
  "cpc": ["G01S7/4863", "G01S17/894"],
  "pdfUrl": "https://patentimages.storage.googleapis.com/c0/d5/f7/86ad5b42759506/US10000000.pdf",
  "googlePatentsUrl": "https://patents.google.com/patent/US10000000B2/en",
  "mode": "details"
}

🗂 Key output fields

Field	When
patentId, title, assignee, inventors[], dates	Always
snippet	Search / assignee / inventor
abstract, description, claims[], cpc[], pdfUrl	Deep / details
citationsBackward[], citationsForward[]	Deep / details
familyId, familyMembers[]	Deep / family
expirationDate, legalEvents[]	Details
citationHop, citationSeed	citationNetwork
googlePatentsUrl, mode, scrapedAt	Always

❓ FAQ

Can I paste a patents.google.com URL?
Yes. Search, assignee, inventor, and /patent/… URLs are classified and merged with the other fields.

How do I get claims and citations?
Set enrichmentDepth to deep, or use mode: "details" with patentIds.

How do I follow citations?
mode: "citationNetwork" with seed IDs, then citationDirection and citationDepth. Cap with maxItems.

Does this need an API key?
No. Search uses Google's public xhr endpoint; details are public HTML.

Coming from another Google Patents scraper?
This one accepts start URLs, multiple queries/assignees/inventors, a CPC filter, full description, and expiry — fields the current Store leader leaves out.

💬 Support

Open an issue on the actor page or email via the Store profile. Include the run ID and the input you used.

🛠 Additional services

Custom fields, scheduled landscaping into a warehouse, or a private build for a filing workflow — say what you need on the Issues tab.

🔎 Explore more scrapers

Sibling IP actor: USPTO Trademark Search. Other actors: memo23 on Apify.

🤖 For AI Agents & LLM Apps

Actor: memo23/google-patents-scraper (slug after first publish).

Purpose: Search Google Patents and return one structured publication record per row.

Minimal input:

{
  "mode": "search",
  "searchQueries": ["lidar perception"],
  "maxItems": 25
}

For a known ID: "mode": "details", "patentIds": ["US10000000B2"].

Behaviors an agent should know:

• One mode per run. Arrays (searchQueries, patentIds, assigneeNames, inventorNames) each run as their own query and merge.
• Always set maxItems. Citation hops grow fast.
• enrichmentDepth: "deep" costs more and fetches the detail page per hit.
• startUrls accept patents.google.com search and detail URLs.
• Pay-per-event — see the Pricing tab.

⚠️ Disclaimer

This Actor reads publicly available patent publications on Google Patents for research, landscaping, and business-analysis use. It is an independent tool and is not affiliated with, endorsed by, or connected to Google, the USPTO, the EPO, WIPO, or any patent office.

The output is not legal advice. Patentability, freedom-to-operate, and validity calls belong to a qualified attorney. Use the data in line with applicable law and Google's terms. The Actor does not store scraped data; rows go to your Apify dataset.

SEO Keywords

Google Patents scraper, Google Patents API, patent search scraper, USPTO patent scraper, EPO patent scraper, WIPO patent search, patent citation network, patent family scraper, CPC patent search, assignee patent portfolio, inventor patent search, prior art search tool, patent claims scraper, patent landscaping, IP intelligence scraper, Google Patents URL scraper, patent JSON export, patent CSV, patent MCP, patent expiry date