jev-decisions
Hermes plugin that asks Jev to review risky plans, check claims against sources, and judge whether a task is really finished.
Overview
Jev Decisions Plugin for Hermes (and other AI agents) gives Hermes extra tools for reviewing a plan, checking an answer against its sources, and assessing whether a task is actually done. You ask for a review in normal conversation and your usual model keeps doing the work. A typical case: Hermes says a document is updated, and the plugin makes explicit whether it opened and checked the saved file or merely received a success response.
It includes 25 prepared reviews: whether a file change or service restart has permission, recovery and a follow-up check; whether sources support a report's claims; whether a draft message is ready, sensitive or makes unintended commitments; whether a proposed memory is useful, sensitive or conflicting; and whether a purchase or next step fits requirements. A local decision journal records observations and outcomes. Reviews are advice only, automatic reviews are off by default, and installing it does not block commands or replace approval settings.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Does this plan include permission, a recovery path and a check afterward? | — |
| Yes / no | Do the supplied sources support the claims? | — |
| Yes / no | Is this message ready to send without sensitive content or unintended commitments? | — |
| Yes / no | Is the task really finished given the evidence?(inferred) | — |
How it works
- Ask Hermes for a specific review in conversation
- Hermes gathers evidence and calls a prepared Jev review
- Jev returns typed judgements
- Hermes reports advice and logs the decision to the local journal
Strengths
- 25 ready-made reviews without learning formats
- Separates claimed success from verified success
- Local decision journal for follow-up
Caveats
- Advice only; does not block dangerous commands
- Automatic reviews off by default
- Hermes must still gather the evidence
Getting started
Follow the Install section of the README, then ask Hermes to use the appropriate Jev review.At a glance
- Type
- Plugin
- Status
- beta
- Runtime
- Hermes agent plugin
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Jev Decisions Plugin for Hermes (and other AI Agents): tool risk reviews, human approval recommendations, evidence checks, and a local decision journal.
logicrw/awesome-jev-projects
Frequently asked questions
What does Jev review?
Plans for risky changes, claims against sources, draft messages, proposed memories, choices and task completion.
Will it stop dangerous commands?
No. Reviews are advice and do not replace approval settings.
Does it work outside Hermes?
The project describes itself as for Hermes and other AI agents; an integration guide is included.
Related projects
Asynchronous System-1 supervisory plugin for Hermes Agent with typed Jev decisions, ranking, verification and an opt-in tool gate.
ProjectbetaPythonAuditable Zig terminal agent pairing Codex for coding with Jev typed preflight and postflight checks under deterministic policy gates.
CLIproductionZigClaude Code plugin with nine Jev-judged gates for rules, scope, intent, completion and claims that escalate but never approve.
HookbetaTypeScriptClaude Code plugin that moves CLAUDE.md rules into hooks, denying rule-breaking edits and unasked deploys via Jev probabilities.
ProjectbetaTypeScriptRisk gate for AI coding agents: Jev decides at five layers and Claude Code hooks enforce deny or ask deterministically.
ProjectbetaJavaScriptJev-powered adversarial or advisory reviewer for the omp coding agent, plus a typesafe_ask tool, plan/stop gates and context compaction.
ProjectbetaTypeScript