Zhihuiya Patent Bibliography

This skill guides you on how to query patent bibliography (bibliographic) data from the Zhihuiya patent database, helping users retrieve detailed metadata for specific patents.

Core Concepts

Patent bibliography data (also called bibliographic data) is the structured metadata associated with a patent document. It includes the patent title, applicants, inventors, classification codes, priority claims, cited references, abstracts, and more. This tool allows querying by patent ID or publication number, returning comprehensive bibliographic records for a single patent per request.

Patent types: The patentType field indicates the type of patent document:

• APPLICATION -- Invention application (published but not yet granted)
• PATENT -- Granted invention patent
• UTILITY -- Utility model
• DESIGN -- Design patent

Data Fields

Invocation

• API Endpoint: POST /zhihuiya/bibliography (full parameters/response/error codes in references/api.md)
• Python Script: python scripts/patentdetailedbibliography.py '<JSON params>' [--inline]
• Cost Constraints: This tool consumes credits. The same parameter combination defaults to a single call per session. The script includes a 24-hour local cache. Do not automatically retry with different keywords, pagination, or modified parameters on failure or empty results; inform the user that additional costs will be incurred before continuing to search. Single Patent Limit: This endpoint consumes many credits. Each call can only pass one patent; if you need to check multiple patents, you must obtain explicit user consent and make separate requests.

Output Strategy (default script behavior):

• Always write the full response to <cwd>/nexscope/<YYYY-MM-DD>/<session>/data/nexscope-zhihuiya-bibliography-<timestamp>.json (<cwd> is the working directory at script execution time, which in Claude Code is the current project directory; <session> is taken from the SESSION_ID environment variable, auto-grouped by user task; writing to /tmp is forbidden -- error if the current directory is not writable)
• Response body <= 8 KB: print the full JSON to stdout after writing to disk
• Response body > 8 KB: print only a summary to stdout after writing to disk (top-level fields, common counts like total/costToken, length of the largest list field + first 3 samples)
• Add --inline to force full output to stdout (still writes to disk)

Data Reading Tip: Check the summary first to determine if it is sufficient; when specific fields are needed, prefer using jq or ConvertFrom-Json to extract from the saved JSON file on demand to avoid loading the entire JSON into context.

Authentication

Set NEXSCOPEAPIKEY. Visit https://www.nexscope.ai/help/skills-external-access?co-from=skillNS to manage credits.

Parameter Guide

The tool accepts two parameters. At least one must be provided; if both are supplied, patentId takes priority.

Tips for Identifying Input Type

• If the user provides something like US10123456B2, CN112345678A, EP3456789B1, or WO2023123456A1, treat it as a publication number and use patentNumber.
• If the user provides a purely numeric or opaque identifier that does not match standard publication number patterns, treat it as a patent ID and use patentId.
• Only one patent may be passed per request. If the user has multiple patents, obtain explicit consent and make a separate call for each.

Usage Examples

1. Look up a single patent by publication number

User: "Show me the bibliography for US10123456B2"
Action: Call with patentNumber = "US10123456B2"

2. Look up a patent by internal ID

User: "Query bibliography for patent ID 8fa3b2c1-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
Action: Call with patentId = "8fa3b2c1-xxxx-xxxx-xxxx-xxxxxxxxxxxx"

Display Rules

1. Present data clearly: Show query results in well-structured tables or organized sections. For each patent, highlight the most commonly needed fields: title, applicants/assignees, inventors, filing/publication dates, classification codes, and abstract.
2. Respect the query scope: Only display the fields the user asked about. If they asked for "inventors", do not dump the entire bibliography unless requested.
3. Patent type labels: Translate patentType codes into human-readable labels (APPLICATION = Invention Application, PATENT = Granted Invention, UTILITY = Utility Model, DESIGN = Design Patent).
4. Single-patent results: Results contain a single patent's data; present it clearly.
5. Error handling: When a query returns an error or empty results, explain clearly and suggest the user verify their patent ID or publication number.
6. No subjective analysis: Present factual bibliographic data without speculative legal or commercial interpretations.

User Expression & Scenario Quick Reference

Applicable -- Patent bibliography / metadata lookups:

Not applicable -- Needs beyond patent bibliography:

• Full-text patent search by keyword or semantic query
• Patent landscape / analytics reports
• Patent valuation or legal status tracking
• Freedom-to-operate or infringement analysis
• Patent family tree exploration (unless specific publication numbers are given)

Boundary judgment: When users say "find patents about X" or "search for patents in field Y", that is a patent search task, not a bibliography lookup. This skill only applies when the user already has a patent ID or publication number and wants to retrieve its metadata.