Jev-chat-assistant
Windows desktop chat assistant that OCRs a selected chat region, has Jev analyse the conversation, and ranks drafted replies for copying.
Overview
Jev Chat Assistant is a Windows PC redevelopment, documented in Chinese with several translations, of the judgment question bank and analysis flow from an open-source jev-chat project. You box-select a chat region on the desktop, local OCR reads the visible text (or you paste text with me/other speaker prefixes), and Jev analyses the conversation; optionally a configured reply model generates three candidate replies, which Jev then ranks. Candidates are only for viewing and copying, never filled in or sent automatically.
Version 1.4.0 adds proofreading before analysis, so users can edit message text, speakers and order before any call, reply purposes such as polite refusal, clarifying a misunderstanding, ending a topic or de-escalating, and rewrite options for candidates. Earlier versions added full-context analysis up to 20,000 characters without silent truncation, supplementary prior context and background, window-following selections, on-demand capture only, cancellation with staged retries and task isolation. A transaction-content check covers the full input. It needs Windows 10/11, WebView2, a Jev key and optionally a reply-model key, and is MIT licensed.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | What is the other person's intent in this conversation?(inferred) | — |
| Choice | Which of three drafted replies best fits the chosen purpose? | — |
How it works
- Capture a chat region by OCR or paste text
- Proofread messages and speakers
- Jev analyses the conversation
- Reply model drafts three candidates; Jev ranks them
- User copies a reply manually
Strengths
- Never sends or fills automatically
- Proofreading step before any API call
- Clear limits instead of silent truncation
Caveats
- Windows only
- Chat content is sent to the configured APIs
Getting started
Run `windows\install.ps1` then `windows\start.ps1` in PowerShell and enter a Jev key in settings.At a glance
- Type
- Project
- Status
- beta
- Language
- Python, TypeScript
- Runtime
- Windows desktop app (WebView2)
- Has code
- yes
- License
- MIT
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-projects
Frequently asked questions
What does Jev decide?
Analysis of the conversation and the ranking of drafted reply candidates.
Does it send replies?
No, candidates are only for viewing and copying.
Can I edit the captured text first?
Yes, version 1.4.0 adds a proofreading step before analysis.
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.
AppbetaSwiftmacOS 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.
HookbetaPythonWeChat reply assistant fork: DeepSeek drafts three replies, Jev pre-judges, reviews and ranks them; auto-send is cancellable.
ProjectbetaPython