User repository on GitHub
微信聊天里的言外之意助手：结合上下文展示情绪概率、潜台词与沟通建议的 Android LSPosed / Xposed 模块。
- Forks
- 14
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Tools & CLIs
Yanwai
Android WeChat assistant that analyzes visible text and voice transcripts for emotion and intent, then drafts分条 reply options and topic openers you copy and send yourself.
- Stars
- 50
- Language
- Kotlin
- Published by
- Community
- Author
- YIRC99
- Last verified
- Oct 2026
Works forJev for Classification
Used inJev for Software & DevTools
Highlights
- On-message emotion probabilities with intent and scenario judgments plus optional LLM intent interpretation.
- Voice handled by native transcription reuse with transcripts kept for checking, never tone analysis.
- Reply drafts split into short selectable messages with visible sources, plus five-topic openers with reasons.
- Per-contact analysis switches, saved identities and background, and separately configured Jev and reply models.
Watch out
Suggestions are reading aids, not proof of intent, and cannot make decisions for you. Long, quoted, image, video, or failed-transcription inputs are skipped or marked unknown rather than guessed. Copy reviewed against the repository README on 2026-10-04, not from running it.
Questions about this entry
Short answers drawn from the linked project or page.
No. The project states it never sends messages automatically. Reply and topic outputs must be selected, copied, and pasted into WeChat by you, and generation only runs when you tap generate.
It needs Android 9 or newer plus an LSPosed or Xposed environment scoped to WeChat, then a WeChat restart. Recorded support centers on one tested build, so other builds need real-device verification.
No. Voice goes through WeChat native transcription first and only the text is analyzed. Tone, crying, and speech rate are out of scope, and analysis states when context is unknown.
Target text plus nearby history goes to the model endpoints you configure, with separate Jev and reply-model channels. Keys stay in private app settings and analysis cache is isolated per account and config.
Three. The default is the Jev decision model returning emotion probabilities; a second uses Jev for emotion plus a language model for intent interpretation; a third uses a pure language model that gives a qualitative emotion read and does not fabricate probabilities, and can reuse the reply model. The pure mode never calls Jev and does not silently fall back to it on failure.
It skips rather than guesses. A single message body or voice transcript over a thousand characters is not analyzed at all, on the principle that truncating and then concluding would mislead. If a target voice transcript fails it stops analyzing that message, and failed voices in the context are marked as unknown instead of inventing content.
It reads at most twenty-four earlier messages from both sides, scanning up to eighty page positions and keeping the combined transcript under twelve thousand characters, and it never reads messages sent after the target. The reply and topic features read more, up to one hundred messages capped at forty-eight thousand characters, but never cross into other conversations.
More Jev resources like this
100GitHub stars
Android chat client and SillyTavern-compatible endpoint with prefix-cache fixes, layered memory, plugins, and Jev-assisted screening.
Jev Tools & CLIs#community#jevbest#catalog-import
168GitHub stars
Android and Windows assistant that reads the chat already on your screen through accessibility nodes or OCR, has Jev judge intent, danger level and urgency in one call, ranks three drafted replies, and fills the input box.
Jev Tools & CLIs#community#jevbest#kotlin
Community
Liyucheng1997/332_lab-jev-chatKotlin
Liyucheng1997/332_lab-jev-chatKotlin44GitHub stars
Mac app that turns one short English or Japanese sentence into Ableton Live mixer, transport, clip, note, device, and track edits from a pop-up bar.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
okinaaudio/live-jevPython
okinaaudio/live-jevPython38GitHub stars
Research dashboard for Polymarket BTC five-minute Up/Down rounds that feeds market state to Jev class scores and paper-trades the edge.
Jev Tools & CLIs#community#jevbest#catalog-import
46GitHub stars
Desktop and web live-call coach that sends transcript lines to TypeSafe Jev after each sentence and shows a 0-100 gauge with a suggested next step.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
ZeroGold/call-coach-aiHTML
ZeroGold/call-coach-aiHTML38GitHub stars
Python memory filter that judges every stored memory against the current request with one calibrated yes/no per memory in a single Jev request, keeping everything above a threshold.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
samdotmak/jev-recallTypeScript
samdotmak/jev-recallTypeScriptFrom the Jev guides
Original write-ups on what this entry is for.
The request shape barely changes across surfaces; the surrounding contract does. When to call the API directly and when a gateway, a Worker binding, or a router entry earns its extra hop.
Jev Tools & CLIs#gateway#vercel#cloudflare
Read article
Taking a decision checkpoint out of Python: what the GGUF and browser WebGPU exports preserve, what they change, and the parity evidence the model cards publish.
Jev Tools & CLIs#open-models#local#onnx
Read article
Jev labels a record that already exists. Code moves, flags, or holds it. How the document, mail, and transfer tools keep the action out of the model.
Jev Tools & CLIs#classification#email#security
Read article
From the Jev community
Posts from builders shipping with Jev right now.