Yanwai
Android WeChat assistant showing Jev emotion probabilities and intent readings beside messages; never sends replies.
Overview
Yanwai (言外) is a free, open-source Android assistant for WeChat, documented in Chinese, that requires Android 9+ and an LSPosed or Xposed environment. It reads the visible conversation, including voice messages transcribed to text, and places an analysis card under a message: emotion probabilities (happy, calm, angry, unclear and so on) shown as colored rings, plus an interpretation of intent, what the other person may care about and emotional tendency, using earlier messages and timing as context.
By default the Jev decision model supplies emotion probabilities; users can also combine Jev emotion with LLM intent interpretation, or use an LLM only, which gives textual judgments without fabricated probabilities. Separate features help draft replies and suggest new topics, which users copy and send themselves; the app never sends WeChat messages automatically. Users supply their own API keys.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | What emotion does this message express? | happy · calm · angry · unclear · ... |
| Choice | What is the intent of this message (confirm, ask, refuse, apologize, thank, clarify)? | confirm · ask · refuse · apologize · thank · clarify |
How it works
- Read the visible WeChat conversation (voice transcribed).
- Jev returns emotion probabilities; an LLM optionally interprets intent.
- Show a card under the message.
- Optionally draft replies or topics for the user to copy.
Strengths
- Probabilities rather than invented certainty.
- Never sends messages automatically.
- Flexible Jev, LLM or combined modes.
Caveats
- Requires a rooted or Xposed-capable Android setup.
- Sends conversation text to model providers.
- Emotion readings are suggestions only.
At a glance
- Type
- App
- Status
- beta
- Runtime
- Android app (LSPosed/Xposed module)
- Has code
- yes
Use cases and integrations
How curated lists describe it
Android accessibility app that reads the visible WeChat conversation and shows emotion probabilities, possible subtext and a suggested reply beside a message. Chinese.
MrJev/awesome-jev
Frequently asked questions
What does Jev provide?
Emotion probabilities for messages.
Does it send messages?
No, it only suggests; you copy and send.
What do I need?
Android 9+, LSPosed or Xposed, and your own API keys.
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 DeepSeek Harness plugin that classifies each user message's emotion, urgency and department with Jev and injects it as context.
PluginbetaTypeScriptAndroid chat co-pilot that reads the active chat via Accessibility and returns typed verdicts on intent, emotion and risk; never replies.
AppbetaKotlinZero-dependency Android app that judges the visible WeChat message's intent, emotion, urgency and reply posture with Jev.
AppbetaJavaAndroid WeChat assistant where Jev judges the chat and scores the relationship, and a chat model drafts three replies.
AppbetaKotlinWeChat plugin that classifies a long-pressed message's intent, emotion, urgency and suggested reply posture with Jev.
PluginbetaJava