Jev directory
CommunityJev Tools & CLIs
Lark MBTI Jev
Windows desktop overlay for Feishu/Lark group chats that syncs messages through lark-cli, uses Jev to score four MBTI dimensions, and draws scroll-tracking name badges over the client. MIT licensed, Python 3.11+.
- Stars
- 1
- Reading time
- 2 min
- Published by
- Community
- Added
- Oct 2026
Highlights
- Windows 10/11 desktop overlay for Feishu/Lark group chats, written in Python 3.11+.
- Uses Jev's Choice primitive to score four MBTI dimensions and compare 16 types.
- Syncs chat history through lark-cli into a local SQLite cache via incremental fetching and backfill.
- Sends up to the 200 most recent messages (40,000 characters) per member to the configured Jev service.
- Labels appear in screenshots and screen sharing and are not validated for personality measurement accuracy.
Questions about this entry
Short answers drawn from the linked project or page.
Clone the repository, run install.ps1 with PowerShell, confirm lark-cli is logged in via 'lark-cli auth status --json', then choose a group chat in the control window and click to enable labels. Windows 10/11 and Python 3.11+ are required.
A configured and logged-in lark-cli with permission to read the chosen group, plus either a Jev relay service (JEV_BASE_URL, JEV_CLIENT_CONFIG, or ~/.agents/jev-client.json) or a TypeSafe API key set through TYPESAFE_API_KEY or TYPESAFE_KEY_FILE.
For a given member it sends that person's most recent messages, up to 200 messages and 40,000 characters, together with message timestamps to the service you configure. Member names and account IDs are not attached separately.
Messages are cached in .data/messages.sqlite3, while analysis results and settings including member names, account IDs, and manual nicknames go to .data/profiles.json and .data/settings.json. Cache, keys, and debug directories are listed in .gitignore.
A member's cached result is shown again until they accumulate 11 new text messages in the same group, after which the next display triggers a fresh analysis. New messages are checked roughly once a minute or on refresh.
No. Results reflect only a person's expression style in that group chat, and the project states the tool has not been validated for personality measurement accuracy. A label like INTP 65% is the model's probability in that analysis.
It targets ordinary group chats in the Windows Feishu client. Topic views, complex nicknames, or different client layouts can interfere with detection, and labels stay hidden when a member cannot be identified; screenshot compatibility mode cannot follow scrolling.
More Jev resources like this
A proof-of-concept HTTP service for constrained decisions with a local MLX model: send text or JSON plus typed choice, score, or noul questions and it returns probabilities over the answers you declared — no generated text, no parsed model JSON.
Jev Tools & CLIs#community#github#python
Community
snellingio/system-onePython
snellingio/system-onePythonA local web app where Claude Code runs the hunt and Jev judges every competitor post, surfacing the ones most likely to work for your audience. It adds saved-post analysis, transcription, an Ask AI view and a My Studio planning page.
Jev Tools & CLIs#community#github#python
A local, non-autoregressive security guardrail suite powered by Convai Innovations' Laya model, returning typed decision probabilities for security inspection. It positions itself against Jev on latency and against generative LLMs on hallucination risk.
Jev Tools & CLIs#community#github#security
Go terminal document classifier built on Jev. It copies a document into a local library, extracts its text, and asks Jev five typed questions covering category, sensitivity, urgency and whether action is required, then files the copy using deterministic Go thresholds and SQLite metadata.
Jev Tools & CLIs#cli#classification#local
Communityglamboyosa/docket
Local job-search command center that runs one Brave site: query per platform across 29 ATS job boards and grades each match with TypeSafe's Jev model on relevance, job, location and still-live axes.
Jev Tools & CLIs#local#javascript#classification
CommunityMaksimZinovev/ats-search-tool
A Go service that polls Hacker News every two minutes, evaluates each new story against customised decision trees using Jev, and posts matching posts to a Telegram chat or channel. The decision trees live in a SQLite database and are fully customisable.
Jev Tools & CLIs#github#classification#routing
Communityuint4096/hndelegate
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.