User repository on GitHub
Agent Skill: send closed coding-agent judgments to TypeSafe Jev
- Forks
- 1
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Tools & CLIs
Jev Judgment
An Agent Skill that sends closed coding-agent judgments to TypeSafe Jev and acts on the probabilities, asking for yes/no, choice, and score answers in about 250 ms while the main coding model keeps writing code.
- Stars
- 6
- Language
- Python
- License
- MIT
- Reading time
- 3 min
- Published by
- Community
- Author
- HyunjunJeon
- Last verified
- Sep 2026
About this project
An Agent Skill that sends closed judgments to TypeSafe Jev and acts on the probabilities. The main coding model keeps writing code. Jev answers yes/no, choice, and score questions in 250 ms so the agent guesses less and interrupts the user only when a person must decide. This repository is the skill only. Install it with the skills CLI; it does not depend on any particular agent harness.
Quoted from the project's README. Read the full README.
Used inJev for Software & DevTools
Highlights
- Calls Jev at three points: before a closed user question, before a possibly destructive command, and after a command fails.
- Deletion outside the request, breaking changes, new dependencies, paid or remote actions, and anything the user flagged always go to the user.
- The bundled scripts/jev.py redacts credentials, truncates long strings, and never blocks the agent on a network error.
- It is the skill only and depends on no particular agent harness; npx skills add discovers the skills/<name>/SKILL.md layout.
- Jev's confidence is evidence about the context, not permission to act.
Quickstart
bash
npx skills add HyunjunJeon/jev-judgment -g -a claude-code codex
export TYPESAFE_API_KEY=YOUR_API_KEYWatch out
MIT-licensed. Requires python3 3.9+ (standard library only), network access to api.typesafe.ai, and a TYPESAFE_API_KEY; without a key the bundled script fails open and the agent continues as if the skill were not installed.
Questions about this entry
Short answers drawn from the linked project or page.
Something runs. The skill ships scripts/jev.py, which POSTs {state, questions} to api.typesafe.ai/v1/systemone at three points: before a closed user question, before a possibly destructive command, and after a command fails. It targets yes/no, choice and score answers in roughly 250 ms while the main coding model keeps writing code.
No. The bundled script never blocks the agent on a network error, and with no TYPESAFE_API_KEY in the environment or in <cwd>/.env it fails open, so the agent carries on as if the skill were not installed. Deletion outside the request, breaking changes, new dependencies and paid or remote actions still route to you regardless of what Jev says.
The bundled scripts/jev.py redacts credentials and truncates long strings before the request is sent. The precise redaction list, the per-dimension thresholds and the question templates are not in the README; they live in references/api.md and references/questions.md under skills/jev-judgment/, so read those before trusting it on a sensitive repo.
No, and the skill says so itself: Jev's confidence is evidence about the context, not permission to act. Its own rule sends deletion outside the request, breaking changes, new dependencies, paid or remote actions, and anything the user flagged to the person. Useful as a cheap prior, useless as an authorization check.
Add the skill with `npx skills add HyunjunJeon/jev-judgment`, or target agents directly with the -g and -a flags. The repo is the skill only and depends on no particular agent harness. Running it needs python3 3.9 or newer using only the standard library, network access to api.typesafe.ai, and a TYPESAFE_API_KEY minted at console.typesafe.ai, set in the environment or in <cwd>/.env. MIT-licensed.
Three distinct checks. Before a closed user question, it asks whether the conversation already determines the answer and proceeds and says so if it does, otherwise it asks you. Before a command that may destroy data, leak secrets, or reach beyond the task, it stops and confirms once any dimension crosses its threshold. After a command fails, it classifies the failure and checks the output for secrets before any retry.
A skills/jev-judgment directory holding SKILL.md, scripts/jev.py, and references/questions.md plus references/api.md, alongside LICENSE and README.md with a Korean translation. The skills/<name>/SKILL.md layout is exactly what `npx skills add owner/repo` discovers. The full protocol, the per-dimension thresholds and the question templates live in those reference files rather than in the README.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
YouTube: Ray Amjad puts Jev inside an agentic coding loop and looks at what it costs to run.
YouTube: RoboNuggets puts Jev to work inside Claude Code.
X: Decision outputs in Grep
Hacker News: jevmem keeps Claude Code project memory by asking Jev which turns are worth saving.
avinashjetwani61 pointsJevmem – automatic project memory for Claude Code, built on JevRead the thread on Hacker NewsMore Jev resources like this
A collection of 26 production-ready agent skills for Claude Code, Cursor, Kiro, Windsurf, and OpenCode; four skills call TypeSafe Jev for calibrated Score and Noul judgments and fall back to heuristics when it is unavailable.
Jev Tools & CLIs#community#python#skills
Community
fabricioctelles/skillsPython
fabricioctelles/skillsPythonClaude Code, Codex, and pi plugin plus MCP server and npm library that hands steps needing no text output to Jev: one batched call answers typed questions about a state, and an opt-in PreToolUse gate can only deny or ask.
Jev Tools & CLIs#community#javascript#claude-code
Community
shitianfang/jev-useJavaScript
shitianfang/jev-useJavaScriptA Stop hook that keeps coding agents from quitting early: rules written in plain language are scored by Jev in about 0.7 seconds, and a violated rule sends the agent back to work instead of letting it stop.
Jev Tools & CLIs#community#python#claude-code
Community
noplan-inc/limpetPython
noplan-inc/limpetPythonA collection of Jev demos, workflows, and agent skills for coding agents: five installable skills and 108 scenarios, installed by pointing Codex, Claude Code, or OpenCode at an agent prompt.
Jev Tools & CLIs#community#python#skills
Community
wuyoscar/jev-skillPython
wuyoscar/jev-skillPythonA local-first MCP server that gives Claude Code, Codex, Cursor, and OpenCode structured quality scores from Jev across correctness, complexity, tests, security, and other dimensions while the agent writes code.
Jev Tools & CLIs#community#typescript#mcp
Community
NiazMorshed2007/jev-reviewTypeScript
NiazMorshed2007/jev-reviewTypeScriptA Jev-guided context compaction library for coding agents: instead of summarizing old context, Jev decides which tool calls and results still matter while user and assistant text is kept verbatim.
Jev Tools & CLIs#community#typescript#compaction
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.