jevvy
Coding-agent plugin that has Jev auto-approve shell permission requests it judges harmless and abstains on anything uncertain.
Overview
Jevvy is a collection of Jev-powered plugins for coding agents, starting with Jevvy Permissions. In Claude Code and OpenCode, the agent's own permission policy is evaluated first: existing allow and deny decisions stay final and configured ask rules still prompt a human. For other unresolved shell approval requests, Jevvy reviews the complete command with a set of inquiries and approves only the current action when every inquiry passes. Otherwise it abstains and the agent's normal permission flow continues.
Configuration lives in one global JSONC file that project repositories cannot override, with a schema for editor validation. Providers include TypeSafe directly and OpenCode Zen, among others. The README notes a Claude Code limitation: its hook can only see readable ask settings, not managed, command-line or session-only rules. It requires Node 22.19+, and setup is a single npx init command. It is aimed at developers tired of approving obviously harmless commands.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Is this shell command harmless to run without asking? (all inquiries must pass)(inferred) | — |
How it works
- Agent's own permission policy runs first
- Unresolved shell approval requests go to Jevvy
- Jev answers inquiries about the full command
- Approve this action only if every inquiry passes
- Otherwise abstain and continue the normal permission flow
Strengths
- Never overrides existing allow, deny or ask rules
- Abstains rather than denies when unsure
- Global config cannot be overridden by repositories
Caveats
- Auto-approval by a model carries residual risk
- Claude Code hook cannot see all rule sources
- Requires a provider key
Getting started
npx @jevvy/permissions initAt a glance
- Type
- Plugin
- Status
- beta
- Language
- TypeScript
- Runtime
- Claude Code and OpenCode plugin
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Jev-powered plugins for coding agents
logicrw/awesome-jev-projectsJev-powered plugins for coding agents.
hellogumbo/awesome-jevPlugins for coding agents, starting with one that auto-approves shell permission requests it judges harmless and passes everything uncertain to the normal flow.
MrJev/awesome-jev
Frequently asked questions
Can it deny commands?
No. It only approves when every inquiry passes; otherwise the normal permission flow continues.
Does it override my rules?
No, existing allow, deny and ask rules stay final.
Which agents are supported?
Claude Code and OpenCode.
Related projects
Local guard layer for coding agents that blocks dangerous tool calls by deterministic rules, with optional Jev judgments.
CLIbetaTypeScriptSecurity hook giving Codex, Copilot, Gemini, Cursor, pi, OpenCode and ACP an auto-mode-like guard via three Jev questions.
LibrarybetaTypeScriptPre-execution hook for coding agents that lets shell commands run, asks a human, or blocks them, with optional Jev classification.
ProjectbetaJavaScriptClaude Code plugin adding Jev judgments to Bash, Write and Edit calls and their outputs; a semantic guardrail, not a sandbox.
PluginbetaEffect-based safety gate for coding-agent shell commands: fast rules first, then Jev decides allow/deny plus eleven risk checks.
ProjectbetaTypeScriptAgent-ergonomic Jev CLI and PreToolUse hook that scores shell commands for hazards and screens fetched text for injection.
PluginbetaTypeScript