jev-chat-windows-deepseek-jev
WeChat reply assistant fork: DeepSeek drafts three replies, Jev pre-judges, reviews and ranks them; auto-send is cancellable.
Overview
This is a public redevelopment of jev-chat-windows, documented in Chinese and not an official release. It recognises the current WeChat chat locally, uses DeepSeek's official API to generate three candidate replies, and optionally calls TypeSafe's official /v1/systemone endpoint for structured judgment, quality review and ranking. The pipeline is Jev pre-judgment, then DeepSeek generation, then Jev review and ranking; if all candidates fail review it rewrites once, and if they fail again auto-send is blocked. Jev never writes replies.
A judgment card shows danger level, true intent with confidence, what the other person needs and a suggested action. Auto-send on Windows has its own switch, off by default, with a 3-second cancellable countdown; the macOS Apple Silicon preview disables it and has not passed real WeChat acceptance. The README warns of a known v1.1.0 bug where fill-in may target the wrong conversation after layout changes, recommending keeping auto-send off and using copy. It requires a DeepSeek key, optionally a TypeSafe key, and Windows 10 1903+ or 11 with WeChat 4.x, and has a custom licence.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Score | How dangerous is this conversation? | — |
| Choice | What is the other person's true intent? | — |
| Yes / no | Does this candidate reply pass quality review?(inferred) | yes · no |
| Choice | Which candidate ranks highest? | — |
How it works
- OCR the visible WeChat chat
- Jev pre-judges danger, intent and needs
- DeepSeek drafts three replies
- Jev reviews and ranks; rewrite once if all fail
- User fills or copies; optional 3-second auto-send
Strengths
- Quality gate blocks auto-send on failing candidates
- Clear judgment card
- Auto-send off by default and cancellable
Caveats
- Known bug may fill the wrong conversation
- Mac build is an unverified preview
- Custom licence
Getting started
Download the release zip, run jev-chat-windows.exe, enter a DeepSeek key and optionally a TypeSafe key.At a glance
- Type
- Project
- Status
- beta
- Language
- Python
- Runtime
- Windows desktop overlay (Mac preview)
- Has code
- yes
- License
- Custom license
Use cases and integrations
How curated lists describe it
This project integrates Jev to provide structured decisions for its workflow. See the repository for implementation details.
logicrw/awesome-jev-projectsWindows 微信回复助手：DeepSeek 官方生成话术，TypeSafe JEV 官方判断排序，支持可取消的 3 秒自动发送。
hellogumbo/awesome-jev
Frequently asked questions
What does Jev do here?
Pre-judges the conversation, reviews drafted replies and ranks them; DeepSeek writes the text.
Is auto-send safe?
It is off by default with a cancellable countdown, but a known bug means you should keep it off.
Is it the official jev-chat-windows?
No, it is a separate redevelopment.
Related projects
- jev-chat-windows★ 802
Windows WeChat reply assistant: local offline OCR reads the chat, Jev judges intent and tension and ranks three drafted replies to fill in.
ProjectproductionPython AstrBot plugin where Jev decides whether the bot should join a group conversation; the main model still writes the reply.
ProjectproductionPythoniOS custom keyboard that shows a copied message's intent, risk level and candidate replies inside any chat app, filling but never sending.
AppbetaSwiftWindows desktop chat assistant that OCRs a selected chat region, has Jev analyse the conversation, and ranks drafted replies for copying.
ProjectbetaPython, TypeScriptmacOS chat overlay that reads the current chat window and shows the incoming message's intent, risk and candidate replies, read-only.
ProjectbetaPythonUnofficial fork of jev-chat-windows: WeChat overlay with local OCR and drafted replies, adding optional auto-send and a knowledge base.
HookbetaPython