Skip to main content
Glama
patent-intelligence-mcp-server

patent-intelligence-mcp-server

by AiAgentKarl

Patent Intelligence MCP Server

PyPI version License: MIT Python 3.10+ MCP

MCP server that gives AI agents access to US patent data. 16 tools for patent search, citation networks, technology trend analysis, IPC/CPC classification, patent families, claims extraction, top holder rankings, portfolio comparison, and strategic landscape analysis — all through the free USPTO PatentsView API.

Features

Tool

Description

search_patents

Search patents by keyword with optional date range filter

get_patent_details

Full patent details: inventors, assignees, CPC classifications

search_by_inventor

Find all patents by a specific inventor

search_by_assignee

Find all patents owned by a company or organization

get_patent_citations

Citation networks with impact scoring

analyze_technology_trends

Patent filing trends over time with growth rates

search_by_cpc

Search by CPC classification code (e.g. G06N for AI/ML)

search_recent_patents

Find newest patents with optional category filter (pharma, AI, etc.)

compare_portfolios

Head-to-head comparison of two companies' patent portfolios

get_patent_landscape

Technology landscape: trends, top players, CPC + geographic distribution

search_by_classification

Search by IPC/CPC code at subgroup level for precise research

get_patent_family

Find related patents (continuations, divisionals) in the same family

get_top_patent_holders

Ranking of companies by patent count, filterable by sector

analyze_patent_landscape

Strategic landscape analysis with insights and recommendations

get_patent_claims

Extract patent claims (the legally binding part)

What's New in v0.3.0

  • 5 new tools: Classification search (IPC/CPC subgroup), patent families, top holder rankings, strategic landscape analysis, claims extraction

  • Enhanced search_recent_patents: Category filter (pharma, AI, telecom, energy, etc.)

  • Enhanced get_patent_landscape: Now includes CPC distribution and geographic distribution

  • Enhanced get_patent_details: Suggests related tools (citations, claims, family)

  • Strategic insights: analyze_patent_landscape generates actionable recommendations

  • 10 sector presets: pharma, tech, semiconductor, telecom, automotive, energy, biotech, AI, medical, aerospace

What's New in v0.2.0

  • 4 new tools: CPC search, recent patents, portfolio comparison, technology landscape

  • Parallel API requests: Trend analysis and patent details are now 3-5x faster

  • In-memory caching: Repeated queries return instantly (10-minute TTL)

  • Date range filters: Search patents within specific time periods

  • Impact scoring: Citation analysis now includes numerical impact scores

  • Better error messages: Helpful hints when no results are found

Related MCP server: Patent by Ouroboros Apps

Installation

pip install patent-intelligence-mcp-server

Quick Start

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "patents": {
      "command": "patent-server"
    }
  }
}

Claude Code CLI

claude mcp add patents -- patent-server

Direct Usage

patent-server

Example Queries

Once connected, your AI agent can answer questions like:

Basic Search:

  • "Search for recent patents about quantum computing"

  • "Show me details for patent US-11234567"

  • "What patents has Elon Musk filed?"

  • "Find all Tesla patents"

Advanced Analysis:

  • "Compare Apple's and Samsung's patent portfolios over 5 years"

  • "Show me the technology landscape for solid state batteries"

  • "Find patents in CPC class G06N (machine learning)"

  • "Show the newest patents about mRNA vaccines from the last 6 months"

  • "Analyze patent trends in CRISPR technology over 10 years"

  • "How does Google's patent portfolio compare to Microsoft's?"

New in v0.3.0:

  • "Search patents in IPC classification H04L9/32 (cryptographic authentication)"

  • "Find the patent family for patent 11234567"

  • "Who are the top patent holders in the pharma sector?"

  • "Give me a strategic landscape analysis for quantum computing"

  • "Extract the claims for patent 11234567"

  • "Show me recent AI patents from the last 30 days"

CPC Classification Codes

Common codes for the search_by_cpc tool:

Code

Area

G06N

Machine Learning / AI

G06F

Digital Data Processing

H01L

Semiconductor Devices

H04

Telecommunications

A61K

Pharmaceuticals

C12N

Biotechnology

B60

Vehicles

F03D

Wind Turbines

API Source

USPTO PatentsView API — Free, no API key required. Covers all US patents with full metadata, inventors, assignees, citations, and CPC classifications.

Configuration

Optional environment variables:

Variable

Default

Description

PATENT_HTTP_TIMEOUT

30

HTTP request timeout in seconds

PATENT_DEFAULT_LIMIT

10

Default number of results

PATENT_MAX_LIMIT

50

Maximum results per query

PATENT_CACHE_ENABLED

true

Enable/disable response caching

PATENT_CACHE_TTL

600

Cache time-to-live in seconds

Performance

  • Parallel requests: Patent details fetch 4 data sources simultaneously instead of sequentially

  • Trend analysis: Year-by-year queries run in parallel (5-year analysis: ~1 request time instead of ~5)

  • Landscape analysis: 5 parallel API calls for comprehensive results in a single request cycle

  • In-memory caching: Repeated queries return from cache (256 entries, 10-minute TTL)

  • Patent families: Parallel citation + base patent lookups

Development

git clone https://github.com/AiAgentKarl/patent-intelligence-mcp-server.git
cd patent-intelligence-mcp-server
pip install -e .
patent-server

License

MIT

Available Tools

15 tools
analyze_patent_landscapeA

Umfassende Landschaftsanalyse für einen Technologiebereich.

Analysiert parallel: Top-Assignees, Anmeldetrends pro Jahr, Top-IPC/CPC-Codes, geografische Verteilung und Wettbewerbsintensität. Gibt eine strukturierte Analyse zurück — ideal für Reports und strategische Entscheidungen.

Unterschied zu get_patent_landscape: Fokussiert auf strukturierte Analyse-Insights statt Rohdaten. Gibt Handlungsempfehlungen.

Args: query: Technologie-Suchbegriff, z.B. "quantum computing", "CRISPR gene editing" years: Analysezeitraum in Jahren (1-20, Standard: 5)

Inputs: query (required), years.

compare_portfoliosA

Vergleicht die Patent-Portfolios zweier Unternehmen.

Zeigt Gesamtzahl, Wachstumstrend und neueste Patente beider Firmen im direkten Vergleich.

Args: company_a: Erste Firma, z.B. "Apple" company_b: Zweite Firma, z.B. "Samsung" years: Vergleichszeitraum in Jahren (1-10, Standard: 5)

Inputs: company_a (required), company_b (required), years.

get_patent_citationsA

Holt das Zitationsnetzwerk eines Patents.

Zeigt welche Patente dieses Patent zitiert und welche anderen Patente dieses Patent zitieren. Berechnet einen Impact-Score basierend auf Zitationshäufigkeit.

Args: patent_number: Patent-Nummer, z.B. "11234567"

Inputs: patent_number (required).

get_patent_claimsB

Extrahiert die Patentansprüche (Claims) eines Patents.

Claims sind der rechtlich bindende Teil eines Patents und definieren den Schutzumfang. Claim 1 ist typischerweise der breiteste Anspruch (Independent Claim), die folgenden sind abhängige Ansprüche.

Args: patent_number: Patent-Nummer, z.B. "11234567" oder "US-11234567"

Inputs: patent_number (required).

get_patent_detailsA

Holt vollständige Details zu einem Patent.

Gibt Titel, Abstract, Erfinder, Assignees, CPC-Klassifikation, Anmeldedatum und Anzahl Claims zurück.

Args: patent_number: Patent-Nummer, z.B. "11234567" oder "US-11234567"

Inputs: patent_number (required).

get_patent_familyA

Findet verwandte Patente in derselben Patent-Familie.

Zeigt Continuations, Divisionals und verwandte Anmeldungen. Identifiziert Patente desselben Assignees die dieses Patent zitieren oder von ihm zitiert werden — typisch für Patent-Familien.

Args: patent_number: Patent-Nummer, z.B. "11234567" oder "US-11234567"

Inputs: patent_number (required).

get_patent_landscapeA

Erstellt eine umfassende Technologie-Landschaft für einen Bereich.

Kombiniert Trend-Analyse, Top-Player, CPC-Klassifikationsverteilung, geografische Verteilung, Aktivitäts-Heatmap und Wettbewerbsintensität. Ideal für strategische Patent-Recherche, Due Diligence und Marktanalysen.

Args: query: Technologie-Suchbegriff, z.B. "solid state battery", "gene therapy" years: Analysezeitraum in Jahren (1-20, Standard: 5) top_n: Anzahl Top-Unternehmen (1-20, Standard: 10)

Inputs: query (required), years, top_n.

get_top_patent_holdersB

Ranking der Unternehmen nach Patentanzahl.

Zeigt die größten Patentinhaber, optional gefiltert nach Technologie-Sektor. Nützlich für Wettbewerbsanalysen und um die dominanten Player in einem Bereich zu identifizieren.

Verfügbare Sektoren:

  • pharma: Medikamente, Wirkstoffe

  • tech: Software, Computing, Hardware

  • semiconductor: Halbleiter, Chips

  • telecom: Telekommunikation, Netzwerk

  • automotive: Fahrzeuge, Antriebe

  • energy: Energie, Batterien, Solar

  • biotech: Biotechnologie, Gentechnik

  • ai: Künstliche Intelligenz, Machine Learning

  • medical: Medizintechnik, Diagnostik

  • aerospace: Luft- und Raumfahrt

Args: limit: Anzahl Ergebnisse (1-50, Standard: 20) sector: Technologie-Sektor zum Filtern (optional)

Inputs: limit, sector.

search_by_assigneeA

Sucht Patente eines Unternehmens oder einer Organisation.

Durchsucht die Assignee-Datenbank nach Firmennamen.

Args: company_name: Firmenname, z.B. "Google", "Tesla", "IBM" limit: Maximale Anzahl Ergebnisse (1-50, Standard: 10)

Inputs: company_name (required), limit.

search_by_classificationA

Sucht Patente nach IPC/CPC-Klassifikationscode.

Präzisere Suche als search_by_cpc — nutzt Subgroup-Ebene für detaillierte Technologie-Recherchen. Unterstützt alle Ebenen: Section (H), Subsection (H04), Group (H04L), Subgroup (H04L9/32).

Beliebte Codes:

  • H04L: Netzwerktechnik, Datenübertragung

  • A61K: Pharmazeutische Präparate

  • G06N: Computing (ML, AI, Neural Networks)

  • H01L: Halbleiter, integrierte Schaltungen

  • B60W: Fahrzeugsteuerung, autonomes Fahren

  • C12N: Biotechnologie, Mikroorganismen

  • H02J: Energieverteilung, Stromnetze

Args: ipc_code: IPC/CPC-Code, z.B. "H04L" (Netzwerk), "A61K" (Pharma), "G06N20" (ML) limit: Maximale Anzahl Ergebnisse (1-50, Standard: 10)

Inputs: ipc_code (required), limit.

search_by_cpcA

Sucht Patente nach CPC-Klassifikationscode.

CPC (Cooperative Patent Classification) ist das internationale Klassifikationssystem für Patente. Nützlich für gezielte Technologie-Recherchen.

Häufige Codes:

  • A: Lebensnotwendiges (Pharma, Medizin, Landwirtschaft)

  • B: Verfahrenstechnik, Transport

  • C: Chemie, Metallurgie

  • D: Textilien, Papier

  • E: Bauwesen

  • F: Maschinenbau, Waffen

  • G: Physik (G06F=Computing, G06N=ML/AI)

  • H: Elektrotechnik (H01L=Halbleiter, H04=Telekommunikation)

Args: cpc_code: CPC-Code, z.B. "G06N" (ML/AI), "H01L" (Halbleiter), "A61K" (Pharma) limit: Maximale Anzahl Ergebnisse (1-50, Standard: 10)

Inputs: cpc_code (required), limit.

search_by_inventorA

Sucht Patente eines bestimmten Erfinders.

Unterstützt Suche nach vollem Namen oder nur Nachname.

Args: inventor_name: Name des Erfinders, z.B. "Elon Musk" oder "Musk" limit: Maximale Anzahl Ergebnisse (1-50, Standard: 10)

Inputs: inventor_name (required), limit.

search_patentsA

Sucht Patente nach Stichwörtern.

Durchsucht Titel und Abstract von US-Patenten. Optional mit Datumsfilter für gezielte Zeitraum-Suchen.

Args: query: Suchbegriff(e), z.B. "machine learning", "battery electrode" country: Ländercode (aktuell nur "US" unterstützt via PatentsView) limit: Maximale Anzahl Ergebnisse (1-50, Standard: 10) date_from: Startdatum im Format YYYY-MM-DD (optional) date_to: Enddatum im Format YYYY-MM-DD (optional)

Inputs: query (required), country, limit, date_from, +1 more.

search_recent_patentsA

Sucht die neuesten Patente in einem Technologiebereich.

Fokussiert auf kürzlich erteilte Patente. Nützlich um aktuelle Innovationen und neue Entwicklungen zu finden. Optional mit Kategorie-Filter (CPC-Section) für präzisere Ergebnisse.

Args: query: Technologie-Suchbegriff, z.B. "autonomous driving", "mRNA vaccine" days: Zeitraum in Tagen zurück (Standard: 90, Max: 730) limit: Maximale Anzahl Ergebnisse (1-50, Standard: 10) category: CPC-Kategorie zum Filtern, z.B. "pharma", "tech", "energy", "automotive", "telecom" (optional)

Inputs: query (required), days, limit, category.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.8/5.0

    Scored across 15 tools

    Disambiguation5/5

    Each tool has a clearly distinct purpose. The overlapping tools like analyze_patent_landscape and get_patent_landscape are differentiated by structured analysis vs raw data. search_by_classification and search_by_cpc use different classification systems. No two tools perform the same function.

    Naming Consistency4/5

    Tool names follow a mostly consistent verb_noun pattern with underscores. Names like search_by_assignee and search_by_classification use a 'by' preposition, while search_patents and analyze_technology_trends do not, introducing minor inconsistency. Overall pattern is clear.

    Tool Count5/5

    15 tools is ideal for a patent intelligence server. The scope covers searching, analysis, comparison, and detailed patent retrieval without being overwhelming. Each tool adds value and the count matches typical user needs.

    Completeness4/5

    The tool set covers core patent intelligence operations: searching by various criteria, detailed patent data retrieval (claims, citations, family), landscape analysis, and portfolio comparison. Minor gaps like legal status or patent valuation are missing, but the domain is well-served overall.

    Maintenance

    ActivityInactive
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      B
      maintenance
      Patent by Ouroboros Apps (https://ouroborosapps.com). USPTO patent search and prior art lookup: search public patent records, fetch a patent, list citations, and run a prior-art search. Not legal advice. USPTO only. Connect by pasting https://patent-mcp.vercel.app/mcp into Claude, Cursor, Grok, or ChatGPT developer mode (OAuth). Docs: https://ouroborosapps.com/docs/patent
      MIT
    • F
      license
      Not graded
      quality
      A
      maintenance
      Enables AI clients to perform hybrid RAG searches over MPEP/USC/CFR patent regulations and BigQuery searches across 100M+ patents, along with automated claims analysis and diagram generation. It exposes tools for prior art search, 35 USC 112 compliance checks, and patent-style figure creation.
      -
    • F
      license
      Not graded
      quality
      A
      maintenance
      Enables AI clients to search patent rules and statutes with hybrid RAG, hunt prior art across 100M+ patents via BigQuery, run automated application compliance checks, and generate patent-style diagrams. It exposes 20+ tools that any MCP-compatible client can call for patent search, analysis, and figure creation.
      -