An Agent Skill that teaches AI agents (Claude Code, Claude.ai, and other SKILL.md-compatible agents) how to call the PatentFig AI API:
- Generate patent figures from text — USPTO/EPO-style line art as PNG or SVG, with optional reference numerals and reference images
- Vectorize raster drawings to SVG / DXF / vector PDF (line-art redraw or faithful trace)
- AI-upscale images 2×/4× with 300/600 DPI metadata
- Convert to filing-ready TIFF / PDF / PNG
Claude Code — copy into your skills directory:
git clone https://github.com/TopLocalAI/patentfig-skill.git ~/.claude/skills/patentfigOr with the skills CLI:
npx skills add TopLocalAI/patentfig-skill- Create an API key at patentfig.ai/settings/api-keys
- Export it:
export PATENTFIG_API_KEY="pfig_..."Then ask your agent things like:
"Generate an exploded-view patent figure of a coffee grinder with numbered parts, as SVG."
"Vectorize drawing.png to DXF so I can edit it in AutoCAD."
API calls draw from your PatentFig AI credit balance — generation 10 credits, conversion endpoints 20 credits, balance lookup free. Credits are only charged on success. See patentfig.ai/pricing.
- Docs: https://patentfig.ai/docs
- OpenAPI spec: https://patentfig.ai/api/openapi.yaml
- Developers: https://patentfig.ai/developers
- Support: contact@patentfig.ai
MIT © TopLocalAI, LLC. The skill instructions are open source; API usage is governed by the PatentFig AI terms.