1. The Production Bottleneck
Coding agents generate patches and execute terminal commands iteratively. Relying purely on generative models for self-critique leads to blind spots, hallucinated library flags, and catastrophic commands (like git reset --hard or rm -rf). Running full-context reviews on every minor diff adds 10-20 seconds to every edit cycle.
2. Architecture Comparison: Without vs With Jev
Agents may execute destructive shell commands without deterministic policy bounds.
Calling large generative models just to inspect a 5-line diff causes high latency and context bloat.
Free-form LLM interpretation often misdiagnoses compiler errors and cycles needlessly.
Evaluate patch risk, syntax confidence, and scope boundary before applying to disk.
Allow, prompt, or block terminal commands using probability-backed safety thresholds.
Instantly pick the right CLI or testing skill without polluting model system prompts.
3. Execution Micro-Flow
How state flows through the bounded decision layer during live agent execution:
4. Practical Engineering Trade-offs
✓Deploy This Pattern When:
- Coding agents execute in autonomous or semi-autonomous terminal loops
- Diffs and commands must be filtered against project safety rules
- Fast triage of linter/test failure outputs is needed
⚠Avoid Or Re-evaluate When:
- Entire architectural redesigns that require multi-file semantic reasoning
- Assuming a green diff gate substitutes for automated test suites
- Missing file context in the serialized diff payload
Read the Jev Engineering for Coding Agents PDF
The 12-page independent brief explores context selection, model routing, tool disclosure, and command gates in a coding-agent harness. Our reading guide links the PDF, explains its ten ideas, and separates design proposals from measured results.
Open the Jev Engineering PDF guide →Following Codex usage updates? See Tibo's sourced Codex reset tracker →
5. Sourced Community Builds
47 ImplementationsElune Coding Agent Switch
An author demo uses Jev to route work between Claude Code and Codex and review risky tool calls.
Hermes iMessage Bridge to Codex
A developer uses Jev for context finding in a Hermes iMessage bridge to multiple Codex projects.
Lahfir Local Harness Router
Jev routes local coding work among Haiku, Opus sub-agents and external harnesses through a Claude Code hook.
Vlad Terin Jev Browser
A Codex browser adapter in which Codex plans, Jev selects the next action and the browser executes it.
Gavin's Jev Codex Model and Effort Router
A macOS Codex wrapper uses Jev to choose a model and reasoning effort from the task.
Auto-Guard for Coding Agent Tool Calls
Jev scores a proposed tool call and the user's request before code allows, escalates or blocks execution.
Clairvoyance Jev Integration for Mac
Clairvoyance's Mac app adds Jev decisions for files, Deep Search and Code Search across coding agents.
Jev CLI for Claude Code & Codex
A zero-dependency Jev CLI and skill for narrow coding-agent judgments.
A dual-arm robot with Jev in the middle layer
Decisions to Jev, inverse kinematics and physics to code.
Voice control over Fusion 360
Jev decides only one thing: was that a command or chatter?
home-assistant-typesafe
Home Assistant routes what you said through Jev, not an LLM.
Hunch
Code review against rules you write in plain English.
is-malicious
A codebase scanner that helps you not run malicious code.
jev-axi
A shell CLI for agents: pick, rate, check, rank, triage and guard.
Jev Browser Use
A Codex skill: Jev clicks, Codex thinks and checks the result.
jev-cua
A local computer-use fast path for Codex and Waku.
jev-desktop
Jev picks the action inside Codex Computer Use.
Jev PR Labeler
GitHub PR labels chosen by Jev, with size measured as scope, not lines.
jev-pruner
Trims long Bash output with Jev before Claude reads it.
jev-router
A per-turn router that sends easy Claude Code or Codex work to a fast model and hard work to a stronger one.
jev-rules
Jev picks which of your rules apply, so Claude only sees those.
jevgrep
Semantic code search for agents: find behaviour, not strings.
jevwire
A decision layer for agents: MCP server, library and Claude Code plugin.
Magic Jev Ball for code review
Shake it on a pull request and Jev approves or rejects in 200 ms.
pi-jev-compaction
Automatic context clearing for Pi, without losing the conversation.
pi-jev
A Pi extension: Jev picks the file excerpts, a local model writes the code.
Hard-coded rules moved to Jev
Ilias Ism uses Jev instead of regex in aiseotracker, linkdr and genppt.
taste-lint
Catches AI slop in a diff before it ships.
typesafe-mcp
An MCP connector that gives Claude Code, Claude Desktop, and Codex typed Jev judgments.
winnow
A context sieve for Claude Code: Jev judges each tool result before it lands.
Jev Agent Toolkit
A portable coding-agent Skill with an optional Jev MCP bridge.
Dynamic GPT-6 Reasoning Effort in Codex via Jev
vechen uses Jev to switch GPT-6's reasoning effort mid-task inside Codex, reducing Astra costs by 50% (author-reported).
jev-chat-jarvis
A reply co-pilot for phone chat apps: it reads the screen and drafts, you send.
Jev-Compatible
Make an existing SGLang or vLLM deployment answer as a decision service.
jev-chat-windows
A WeChat reply helper for Windows: screenshot, local OCR, three drafts.
jev-rs
A Rust, Jev-compatible /v1/systemone engine, one prefill per judgement.
jev-test-filter
Score every test against a git diff, then emit the runner arguments.
jev-browser-wingman
A bounded browser-action tool for Claude Code and OpenCode on a real logged-in Chrome session.
jev-browse
One-call browser sub-tasks for Claude Code and Codex, with Jev choosing each low-level action.
JEV Fuse
A governed execution layer that turns Jev probabilities into deterministic policy actions.
jev-kit for Claude Code
A Claude Code kit that wires Jev into guards, sub-agent sizing, search, browsing and review.
Jev Subconscious Helper
A lightweight decision layer that filters logs, blocks risky commands and speeds up small coding-agent checks.
CasaJev
A local Jev harness that builds, verifies and reuses narrow tools across agent tasks.
ClearJev — Codex Prompt Model Routing
An independent Codex CLI skill where Jev selects a model for each prompt.
Astra-Ares — Adaptive Reasoning in Codex
An experimental patched Codex CLI that uses Jev to adjust GPT-6 Astra reasoning effort mid-task.
Jev Semantic Code Reading for Claude Code
A Claude Code plugin that uses Jev to locate relevant code and narrow large reads before the agent consumes context.
Claude Code Project Memory with jevmem
Keep useful decisions between coding sessions instead of saving every message.