jev-guard
Safety gate plugin for Claude Code, Codex CLI and Antigravity: local sensitive-file checks, then Jev judges blast radius.
Overview
jev-guard intercepts tool calls in Claude Code, Codex CLI and Antigravity before they run: shell commands, file writes, patches, file reads and directory listings. It first runs sub-millisecond local checks for boundaries and sensitive files, prompting for confirmation on credentials, .env files, SSH, AWS, Kubernetes, npm, PyPI and git credential files, private keys and certificates, including XDG config paths and unexpanded variable references. Remaining calls go to Jev to evaluate blast radius, reversibility and destructive potential.
It maps each agent's tool names (for example Bash, Edit and Write in Claude Code; exec_command and apply_patch in Codex; run_command and write_to_file in Antigravity). MCP tools are not gated by default. The README documents eval and doctor commands and a config file.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Score | What is the blast radius of this tool call?(inferred) | — |
| Yes / no | Is this action irreversible or destructive?(inferred) | — |
How it works
- Hook intercepts a tool call
- Local checks flag sensitive files and boundaries immediately
- Jev evaluates blast radius and reversibility
- Allow, ask or block per policy
Strengths
- Works across three agents
- Sensitive files caught locally before any network call
- Broad credential path coverage
Caveats
- MCP tools not gated by default
- Requires a TypeSafe key for Jev checks
At a glance
- Type
- Plugin
- Status
- beta
- Language
- TypeScript
- Runtime
- Agent hook plugin
- Has code
- yes
- Has tests
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
High-speed, cross-agent safety gate plugin for \*\*Claude Code\*\*, \*\*Codex CLI\*\*, and \*\*Antigravity\*\*.
logicrw/awesome-jev-projects
Frequently asked questions
What does Jev decide?
How destructive and reversible a tool call is.
Which agents are supported?
Claude Code, Codex CLI and Antigravity.
Are MCP tools covered?
Not by default.
Related projects
Claude Code hook where Jev reviews each tool call against risk, authorization and scope questions and a local policy returns allow or deny.
ProjectbetaGoClaude Code mod that answers auto mode's tool-call permission question with eight Jev yes/no checks behind a shell-aware blocklist.
ProjectbetaTypeScriptClaude Code hook that blocks known key formats locally and sends unknown high-entropy strings to Jev only in masked form.
HookbetaPythonTool-call firewall for Claude Code hooks and MCP servers: Jev answers seven risk questions and code maps them to allow, ask or deny.
MCP serverbetaTypeScriptRing Zero: kernel-level runtime security for AI agents, with Jev as an optional check that can only raise severity, never permit.
ServicebetaOpenCode 2 plugin that asks Jev six questions about every shell command (local or remote via FarHand) and prompts unless clearly safe.
MCP serverbetaTypeScript