User repository on GitHub
pre-commit hook: one Jev call judges whether your commit message matches the diff, plus debug leftovers, scope creep, and a secret belt
- Forks
- 2
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Tools & CLIs
Jev Commit
A pre-commit hook that answers five Jev questions about the commit message and staged diff together, warning on mismatches and blocking only on a credential-shaped line.
- Stars
- 16
- Language
- Python
- License
- MIT
- Reading time
- 4 min
- Published by
- Community
- Author
- valentynkit
- Last verified
- Sep 2026
About this project
Your agent writes the code, then writes the commit message about the code. Nothing checks that the two agree. This does, in one call to Jev, before the commit lands. Four cents per thousand commits. It warns and gets out of the way. The one thing it stops is a credential on an added line. Five questions about the message and the staged diff together, in one request. A high number is always the bad direction.
Quoted from the project's README. Read the full README.
Used inJev for Security & Privacy
Highlights
- Five questions in one request flag filler messages, claims absent from the hunks, debug leftovers, unmentioned work, and secret-shaped values.
- Over a 152-commit corpus the median cost was $0.000044 per commit, about four cents per thousand commits.
- Warnings never block; only a credential on an added line stops the commit, and --strict blocks on findings past the threshold.
- Needs pre-commit install --hook-type commit-msg, since a plain pre-commit install will not run it.
- Python 3.10+ with no dependencies, and the hook itself takes 240 ms median, 21 ms of it a loopback call.
Quickstart
bash
pipx install git+https://github.com/valentynkit/jev-commit
printf '#!/bin/sh\nexec jev-commit "$1"\n' > .git/hooks/commit-msg
chmod +x .git/hooks/commit-msgWatch out
MIT-licensed. Needs a TypeSafe API key, and pre-commit refuses to install while core.hooksPath is set, so the hook must be added by hand there.
Questions about this entry
Short answers drawn from the linked project or page.
It lets them through. Every error path fails open: an API error, a timeout past eight seconds, a missing key, or a Ctrl-C exits 0 and the commit proceeds. Exit 20 is reserved for a credential-shaped added line, or a finding past the threshold under --strict, and 2 is a usage error. That is a deliberate trade: a hook that blocks on its own infrastructure stops being a warning.
Partly, and the gaps are stated. Cost is arithmetic over a committed 152-case corpus at list price rather than a billed figure, and the timing row was measured against a local fake on an M-series Mac. The detection rate is not published: the README says make measure prints it from recorded answers and the number will be carried once it has run against a real key.
Likely yes, which is the main friction for test-heavy repos. The belt is regex plus entropy, so it blocks on credential shapes rather than on real credentials. The documented escape hatches are --exclude on that path or a trailing # jev-commit: allow comment on the line, which the belt skips. Thirteen high-precision patterns block; two high-recall ones only report.
Not all of them. Git runs no commit-msg hook for merges or rebases, and --no-verify skips hooks entirely, so those are silent by construction. git add followed by git commit --amend is byte-identical to a plain commit from inside the hook and is judged against the newly staged delta alone; --amend-base opts into HEAD^ and never blocks, falling back to the index on a root commit.
A filler message, a message that contradicts the diff, debug leftovers, unmentioned work, and a secret-shaped value. A high number is always the bad direction. Warnings never block; only a credential-shaped value on an added line stops the commit, and --strict can block on a finding past the threshold.
The hook itself takes 240 ms median, of which 21 ms is the loopback call, measured on an M-series Mac. The rest is a fresh Python process: about 42 ms of interpreter, 117 ms of stdlib imports and 60 ms of git, chunking and the credential belt.
The belt is regex plus entropy, so it blocks on credential shapes rather than real credentials. Use --exclude on a test-fixture path, or add a trailing # jev-commit: allow comment on the line, which the belt skips. Thirteen high-precision patterns block and two high-recall ones only report.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Vercel's fx safety reviewer, measured at 18× faster
Reddit: Early experiments wiring Jev into Pi as a safety layer and model router.
X: Decision outputs in Grep
YouTube: Kubernetes admission control with a decision model, fail-closed.
More Jev resources like this
Pi extension that supervises a coding agent with Jev judgments: it holds risky tool calls, checks writes against project Markdown rules, and feeds most issues back to the agent as a steer instead of interrupting you. The conscience is beta and off by default.
Jev Tools & CLIs#community#typescript#guardrails
Community
DevMortimer/pi-wardenTypeScript
DevMortimer/pi-wardenTypeScriptA CLI that turns Jev into Unix pipeline primitives: is, choose, score, filter, and guard commands return typed verdicts, stable exit codes, and versioned JSON without prompt or parsing glue.
Jev Tools & CLIs#community#python#cli
Community
sharziki/semdecidePython
sharziki/semdecidePythonA security hook for coding agents that asks Jev three typed questions before every tool call and scans tool results for prompt injection, with adapters for Claude Code, Codex, Copilot CLI, Gemini CLI, Cursor, pi, OpenCode and ACP.
Jev Tools & CLIs#community#javascript#agents
Community
leepokai/jev-guardJavaScript
leepokai/jev-guardJavaScriptAgent-ergonomic CLI following the AXI conventions that gives coding agents Jev judgments for blocking risky tool calls, screening fetched content for prompt injection, triaging logs, flagging risky diffs, and ranking many items. Routine commands are decided locally at no cost.
Jev Tools & CLIs#community#typescript#cli
Community
shiftynick/jev-axiTypeScript
shiftynick/jev-axiTypeScriptA single-binary Rust gate that screens staged git diffs for unredacted secrets, prompt-injection attacks, and destructive commands with calibrated Jev probabilities in about 80 ms, invoked as git jev.
Jev Tools & CLIs#community#rust#cli
Community
AkashPriyadarshii/jev-gitRust
AkashPriyadarshii/jev-gitRustA Node CLI that sends source, configuration, build, and CI files to Jev and reports suspicious files and lines with probabilities, as a second opinion before running unfamiliar code.
Jev Tools & CLIs#community#typescript#cli
Community
luantak/is-maliciousTypeScript
luantak/is-maliciousTypeScriptFrom 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.