shipwithjev

Catalog / Agents & browsers

0051GitHub

jev-desktop

Jev picks the action inside Codex Computer Use.

A harness that keeps Codex Computer Use for the screen work and hands the action selection to Jev.

yikangy873-gif/jev-desktopREADME ↗
# Jev Desktop for Codex

[](https://github.com/yikangy873-gif/jev-desktop/actions/workflows/test.yml)

[](LICENSE)

Jev Desktop adds a bounded decision loop to Codex Computer Use for browser tabs and native macOS apps.

- **Codex** understands the goal, prepares text, defines the allowed actions, and verifies the result.
- **TypeSafe Jev** selects an operation and its compatible target in one request.
- **Computer Use** observes the approved interface and performs the actual click, fill, scroll, or shortcut.

It reuses the existing Codex Computer Use runtime. It does not install a second browser controller, call private Computer Use APIs, require Browser Harness, or use OpenRouter to generate field text.

## Why this exists

Returning to the main agent after every UI click adds latency and loses short-term action context. Jev Desktop keeps one bounded session inside the existing Computer Use runtime, while Codex retains control of scope, permissions, prepared values, recovery, and final verification.

The plugin is intentionally not a general unrestricted desktop agent. Every executable control must be explicitly allowed by the calling Codex task.

## How it works

```mermaid
flowchart TD
    U[User goal] --> C[Codex plans and scopes the task]
    C --> F{Deterministic fast path?}
    F -->|Yes| D[Direct connector or CUA action]
    F -->|No| S[Create bounded Jev session]

    subgraph Local[Local machine]
        S --> O[CUA reads a full accessibility snapshot]
        O --> P[Parse indexed elements]
        P --> A[Apply action and safety allowlists]
        A --> R{Sensitive or consequential control?}
        R -->|Yes| H[Return control to Codex]
        R -->|No| Q[Build operation head and compatible target heads]
        Q --> BC[Authenticated loopback bridge]

   

Also filed under Agents & browsers

  1. 0226

    Flight search with Browser Use★

    Breaking: Browser Use + Jev = Ultrafast ⚡ Findings flights took 7s and cost only $0.0039 🤯 > new action space every step > DOM state space > small LLM fallback to type (this video is at 1x speed btw) Built a tiny open source browser agent. try it below ↓

    @gregpr07 · Agents & browsers · ~$0.004 · ~7 s

  2. 0619

    jevmem: automatic project memory for Claude Code

    After every Claude Code message, Jev decides in ~0.3 s whether it holds a decision, rule or bug worth keeping, and jevmem saves it as one line in JEVMEM.md.

    @jetwaniavinash · Agents & browsers · ~0.3 s

  3. 0618

    Playwright + Jev: tests in plain language

    Write Playwright steps like "add the most expensive item to the cart". Jev finds the right buttons and fields, fills forms and checks the result.

    Andrey Popov · Agents & browsers

  4. 0617

    toolgate: a tool-call firewall for agents

    A Claude Code and MCP hook where Jev answers seven yes/no risk questions per tool call to allow, ask or deny. Usage and benchmark numbers are in the repo.

    Jaz · Agents & browsers