YangCazz/CazzPatent

YangCazz★ 0Python最后同步: 2026-08-14

在 GitHub 打开

AI patent disclosure drafting plugin for DeepSeek Harness - 8-stage pipeline, LaTeX to OMML, diagram generation, self-improving memory

README 摘要

English 中文 CazzPatent AI patent disclosure drafting assistant — a DeepSeek Harness plugin that turns technical proposals into submission-ready patent disclosures. What is this? CazzPatent is a DeepSeek Harness plugin that acts as an AI patent agent. It walks a raw technical proposal through an 8-stage pipeline — environment check, material parsing, patent-point mining, prior-art search, drafting + diagram generation, review & iteration, reflection & learning, and multi-format export — producing a disclosure ready for firm review. It ships as two complementary parts: Part Role Install cazz-patent Skill Instruction brain: 8-stage workflow + self-improving memory + 6 Python tools Drop into .dsh/skills/ or customSkillDirs plugin/ Cordis plugin Deterministic tools: 5 schema-validated Python tool wrappers dsh plugin add github:YangCazz/CazzPatent Pipeline Stage 5.5 (Reflect & learn) extracts reusable rules from iteration history, dual-scores them, and regenerates injection fragments — "fix once, never repeat" — so later proposals benefit automatically. Highlights - LaTeX → OMML native equations : 350+ symbols + 96 structural commands — fractions, sub/superscripts, sums, integrals, matric…

在 GitHub 查看完整 README →
Tools / Devaideepseek-harnessdsh-pluginlatexpatent

分类