jevbooks

← All projects

jev-patent-disclosure

hoangngochuong24947-gif/jev-patent-disclosure

Jev gates patent drafting: 300ms fact check before LLM writing.

What it is

A Chinese patent skill suite for R&D teams and patent engineers: mining patent points, scoring fact completeness, judging legal patent type, and generating disclosure/application documents, OA replies and Obsidian patent maps.

How it uses Jev

patent_intake_gate.py runs a Jev System-1 gate on the raw technical brief. Jev returns a factual completeness score and a patent category (invention/utility model/design). If completeness < 1.5 the script exits with code 2 and emits a missing-fact question list; if >= 1.5 it passes to System-2 LLM drafting.

Primitives:scorechoice

Technique worth stealing

Use a fast typed gate to reject under-specified inputs before spending slow LLM drafting compute.

Try it

pip install -r requirements.txt; export OPENROUTER_API_KEY; python3 examples/demo_jev_intake_comparison.py

View on GitHub

judged by Jevjev-1.13.0

Evidence

Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.

  • Jev-centricyes0.88
  • Shows a System One patternyes0.69
  • Handles uncertaintyno0.15
  • Measuredno0.23
  • Runnableyes0.68
  • Worth recommendingno0.34
  • Model replicano0.15
  • Problem scopescore on a 0–2 scale1.04
  • About Jevyes0.93

Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 20 Sept 2026.