Skip to content

Jev Decisions

Hermes plugin that puts an agent's plans, drafts and claimed completions up for review — plan_review, evidence_review, action_verify — plus local approval rules that make no model request.

Stars
3
Language
Python
License
MIT
Published by
Community
Author
bojansandhaus
Last verified
Oct 2026

Jev Decisions Plugin for Hermes (and other AI Agents): tool risk reviews, human approval recommendations, evidence checks, a local decision journal, local decision supervision, and learned corrections.

Forks
0
Last pushed
Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

README · Technical reference · Integration guide · Contributing · MIT license. Jev Decisions Plugin for Hermes (and other AI Agents) Help Hermes check its plans and its work. Review a risky change, catch an unsupported claim, or check what still needs doing. Install · Try it · About the model · FAQ · Technical reference.

Quoted from the project's README. Read the full README.

Highlights

  • 25 prepared reviews cover file changes and service restarts, research claims against the sources read, message drafts, memory writes and purchases.
  • Three levels of use: one requested review, local rules only with no model request, or opt-in automatic observation that adds requests, records and delay.
  • The order is the point — perform the action, collect the evidence, then review; judging a success message without reading the target adds an opinion, not proof.
  • Install is a git clone into the plugin directory plus hermes plugins enable, hermes tools enable and hermes plugins doctor.

Watch out

The plugin states in bold that reviews are advice: installing it does not automatically stop dangerous commands or replace your approval settings, automatic reviews are off by default, and even opt-in automatic observation records reviews without blocking anything. Platform-wise this release uses POSIX journal locking, Linux is what CI tests, and native Windows is unverified with the locking module unavailable there, so use WSL. The evidence remains the agent's job — the README's own framing is that Hermes must open the saved file rather than trust the editing tool's success message, and verify_observation only compares the values it is handed, so a reviewer pointed at the wrong target still sounds confident. Model reviews need a hosted provider key or a local Laya server you run, while the local rules that check approval need and evidence make no model request but cannot interpret the task. Hermes writes the explanation and is told not to present it as Jev's reasoning. Nothing here measures how often a review catches a real gap, the decision journal is local with unknown outcomes left unknown, and the private infrastructure collectors are not in the public package. 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.

More Jev resources like this

Shell gate for coding assistants that judges commands by effect and blast radius, asking Jev twelve questions or a chat model, with fast rules first.
Jev Agents & MCP#community#awesomejev#catalog-import
6GitHub stars
Python helper that wraps TypeSafe Jev for agent arbitration, pre-execution guard, stuck-loop checks, fact verification and intent routing with probabilities and confidence.
Jev Agents & MCP#community#awesomejev#catalog-import
1GitHub stars
An OpenAI-compatible proxy that spends one Jev call per request on injection, harm and PII checks plus a tier label, then routes to a small or large backend from pure code.
Jev Agents & MCP#community#jevbest#catalog-import
51GitHub stars
Local LinkedIn recruiting workspace on Browser Use's Jev Ultrafast: Jev screens titles, drives navigation, and assesses profile excerpts against your brief while a live browser stays visible; matches and evidence export as JSON.
Jev Agents & MCP#community#jevbest#catalog-import
33GitHub stars
Codex Skill plus stdlib-only Python CLI that recommends which installed skill to read next for a task, allowing no-skill or uncertain answers and never executing candidates.
Jev Agents & MCP#community#jevbest#catalog-import
72GitHub stars
MCP server over stdio that gives coding agents fourteen Jev judgment tools, turning evidence plus an enumerable question into auto, review, or escalate actions from calibrated probabilities.
Jev Agents & MCP#community#jevbest#catalog-import
Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai