#3232nd of 36 in Assistants
jev-chat-jarvis
Chat reply assistant that judges intent and risk before drafting replies
- Stars
- 8k
- License
- MIT
- Last commit
- Oct 2026
- Last release
- Oct 2026
- Language
- Kotlin
5,389 on 23 Sept 20267,966 on 21 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Reads a conversation on screen, works out the other person's intent and the risk in the moment, then drafts and ranks replies that fill into the message box. The Android apps in this repo overlay WhatsApp, QQ, Feishu and X via the accessibility service. Separate repos cover Windows, macOS and an iOS keyboard. Requests go only to the model provider you configure with your own key.
Who it is for: People who want AI-drafted replies inside chat apps on phone or desktop
Strengths
- Never sends messages itself; it only fills the input box
- Requests go only to your configured provider with your own key; no vendor server
- Covers Android, Windows, macOS and iOS, MIT-licensed
- Drafts are checked against the intent and risk judgment and ranked
Weaknesses
- Use inside WhatsApp, QQ, Feishu or X may breach their terms and risk bans
- Chinese Android edition needs an ARM64 phone; macOS build is Apple Silicon only
- iOS keyboard has no App Store build; you must compile it in Xcode
- Windows, macOS and iOS code live in other repos; data handling documented there
What it needs
- Needs Model provider API key (user-configured), Android accessibility service and overlay permission
- Models: User-configured model provider via own API key, Local model on macOS
Also in Assistants
See all 36| Rank | Project | Score |
|---|---|---|
| 1 | ai-job-searchClaude Code workflow that searches job boards and drafts tailored CVs and cover letters | 92 out of 100 |
| 2 | OpenClawPersonal AI assistant reachable through Discord, Slack, Telegram, WhatsApp and more | 85 out of 100 |
| 3 | nanobotSelf-hosted personal AI agent reachable from browser, terminal and chat apps | 85 out of 100 |
| #4 | openhumanRust agent harness with desktop app, built-in memory and embeddable library | 85 out of 100 |
| #5 | ZeroClawSingle Rust binary agent runtime with 30+ channels and hardware access | 80 out of 100 |