figure-description

Generates formal patent drawing descriptions and reference numeral indexes from user-provided figures.

Updated Jul 9, 2026
One-click install
npx skills add https://github.com/Lingjie-wang/autoRL --skill figure-description-lingjie-wang
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: figure-description
Source: https://github.com/Lingjie-wang/autoRL/tree/main/Auto-claude-code-research-in-sleep/skills/figure-description
Command: npx skills add https://github.com/Lingjie-wang/autoRL --skill figure-description-lingjie-wang

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Drafting patent applications requires formal drawing descriptions and consistent reference numerals mapped to claim elements, which is tedious and error-prone when done manually across multiple figures. ## Core Features & Use Cases - Figure Discovery and Analysis: Automatically searches the project directory for PNG, JPG, SVG, and PDF figures, then identifies components, connections, and flows in each image. - Reference Numeral Assignment: Assigns numerals using a consistent 100-series convention (FIG. 1 uses 100-199, FIG. 2 uses 200-299) and reuses numerals for components appearing in multiple figures. - Claim Cross-Referencing: Maps every numeral to claim elements and flags missing figures or numerals that the claims require. - Use Case: A patent drafter has three technical diagrams and a CLAIMS.md file; the skill produces figure_descriptions.md with formal CN or US/EP style descriptions plus a complete numeral index table. ## Quick Start Run the figure-description skill on the figures in my patent/figures directory to generate drawing descriptions and a reference numeral index mapped to my claims.

Frequently Asked Questions about figure-description

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

FAQPage Schema
How do I generate patent figure descriptions from technical diagrams?

Place your figures (PNG, JPG, SVG, or PDF) in the project directory and invoke the skill with the figure directory or list. It reads each image, identifies components, assigns reference numerals, and writes formal drawing descriptions to patent/figures/figure_descriptions.md.

How are reference numerals assigned to patent figure components?

Numerals follow a 100-series convention: FIG. 1 uses 100-199, FIG. 2 uses 200-299, and so on. Components appearing in multiple figures keep the same numeral, and each numeral is cross-referenced to the claim element it supports.

Does the skill support both Chinese and US patent drawing description formats?

Yes, it generates CN jurisdiction descriptions in Chinese (图1是...示意图) and US/EP descriptions in English (FIG. 1 is a block diagram illustrating...). Each description is a single formal sentence in a consistent format.

What happens if a claim element has no corresponding figure?

The skill flags gaps during cross-referencing with markers like [MISSING FIGURE: Need a diagram showing the element] or [MISSING NUMERAL]. This ensures every claim element has at least one reference numeral before filing.

Can the skill modify or annotate my original figure files?

No, the skill only reads and describes user-provided figures; it never modifies them. Its outputs are limited to figure_descriptions.md and numeral_index.md in the patent/figures directory.