Patent Research Skill

Patent Research Skill is by GitHub user cyy1018abc-kyy, and the README says it works with Claude Code, Codex CLI and the Codex desktop app. Four Skills share a patentkit core: prior-art search, mining R&D material for patentable points, drafting the application, and auditing the evidence chain. The output is SVG and docx built on your machine, with optional PDF. An optional patent_intel module goes online to query patent and literature data such as CNIPA and Crossref; the author marks the OPS channel as only partly working. What it produces is an internal draft, not a filing, and it isn't legal advice.

Who it's for:R&D teams and patent engineers who want prior-art search and first drafts in order. Readers who don't write code won't get much from it.

Before you start: You need the Skill pack plus the patentkit Python package (Python 3.11 or later). Windows has a one-click installer, or you can install from a shell or PowerShell. The README's POSIX example names wheel version 0.2.0, which doesn't match the required 0.10.0, so check you have the right version. By default the installer stops if the target exists; --force or uninstall deletes the old Skill folder, and a sample self-check runs after install. The README asks you not to upload unpublished material to cloud clients. External patent data services may have quotas or need authentication. The Skills are written mainly in Chinese, some with English references; the scripts are in English. Apache-2.0 license.

Based on the project's public documentation. Not tested by AI Inspiration World.