patsnap-patent-literature-search

Search patents and scientific literature through PatSnap's MCP server using natural language queries.

7|2|Updated May 16, 2026
One-click install
npx skills add https://github.com/reason-machines/mcp-skills --skill patsnap-patent-literature-search-reason-machines
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: patsnap-patent-literature-search
Source: https://github.com/reason-machines/mcp-skills/tree/main/skills/patsnap-patent-literature-search
Command: npx skills add https://github.com/reason-machines/mcp-skills --skill patsnap-patent-literature-search-reason-machines

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Researchers, IP professionals, and R&D teams need to search massive patent and academic literature databases, but traditional Boolean search syntax and fragmented databases make prior art searches, competitive intelligence, and freedom-to-operate analysis slow and error-prone. ## Core Features & Use Cases - Natural Language Patent Search: Query 200M+ patents across 170+ jurisdictions (USPTO, EPO, WIPO) without learning Boolean syntax, with filters for assignee, inventor, IPC class, legal status, jurisdiction, and date range. - Scientific Literature Search: Search 216M+ academic papers with citation-count filtering, and combine with patent results for fusion research workflows. - Detailed Record Retrieval: Fetch full patent records (claims, descriptions, legal status, patent family) or literature metadata as Markdown by URL or publication number. - Use Case: Before filing a patent on a solid-state battery design, search granted US patents by assignee and date range, then fetch the full claims of the top results to assess prior art and freedom to operate. ## Quick Start Ask the AI to search PatSnap for granted US patents on solid state battery lithium metal anodes filed by Toyota since 2020, then fetch the detailed records of the top three results.

Frequently Asked Questions about patsnap-patent-literature-search

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I search patents with natural language instead of Boolean syntax?▼

Use the patsnap_search tool with a plain-language query string and set source to "patent". The server interprets semantic intent, so queries like "solid state battery lithium metal anode" work without Boolean operators, and you can add structured filters for precision.

How to do a prior art search before filing a patent?▼

Run patsnap_search with your invention description, a date_to filter before your filing date, and the target jurisdiction. Then call patsnap_fetch on the top result URLs or publication numbers to review full claims and descriptions of the closest prior art.

Does PatSnap MCP support filtering patents by assignee or IPC class?▼

Yes, patsnap_search accepts a filters object with assignee, inventor, ipc_class, legal_status, jurisdiction, date range, and minimum citations. IPC codes can be broad (H01M10) or specific (H01M10/0562) depending on the precision needed.

Can I use PatSnap patent search with Claude Desktop or Cursor?▼

Yes, it connects as a streamable HTTP MCP server. For Claude Desktop run the claude mcp add command with the PatSnap URL and your API key; for Cursor or Windsurf add the server URL to your MCP configuration JSON file.

Why does my patent search return zero results?▼

Zero results usually come from overly specific queries or restrictive filters. Simplify keywords, widen the date range, remove jurisdiction limits, use broader IPC classes, and verify assignee name spelling before retrying.

What are the API limits for PatSnap patent and literature search?▼

New accounts receive 10,000 free credits, and each search consumes credits based on page size. Reduce page_size, fetch full records only for relevant results, and monitor usage on the Patsnap Open Platform to avoid rate limit errors.