Free tool

Agent config generator

Point your coding agent at a local model gateway. Pick an agent, enter the endpoint, and copy the exact config snippet — the same fields AutoJev writes when it connects that agent.

Runs entirely in your browser. Nothing you type is sent anywhere.

Agent

OpenAI-style base URL ending in /v1. Claude Code and Gemini CLI use the origin without /v1.

Codex: AutoJev does not write a key for the local provider, so this field is ignored.

An AutoJev route such as autojev/fast, or a single model such as openrouter/vendor/model.

Protocol

Codex talks to the gateway with the OpenAI Responses API (wire_api = "responses").

Config to add · Codex

Shown as it would look in a fresh config file. If the file already exists, merge these keys into it — keep your other settings. This agent needs 2 files.

~/.codex/config.toml
model = "autojev/fast"
model_provider = "autojev"
# AutoJev writes the absolute path of ~/.codex/autojev-models.json here
model_catalog_json = "/Users/you/.codex/autojev-models.json"
web_search = "disabled"
 
[model_providers.autojev]
name = "AutoJev"
base_url = "http://127.0.0.1:9527/v1"
wire_api = "responses"
 
[model_providers.autojev.http_headers]
x-autojev-agent = "codex"
~/.codex/autojev-models.json

Model catalog so the model appears in Codex’s picker

{
  "models": [
    {
      "apply_patch_tool_type": "freeform",
      "base_instructions": "You are a coding assistant. Use the available tools to complete the user's task.",
      "context_window": null,
      "default_reasoning_level": "medium",
      "default_verbosity": null,
      "description": "AutoJev · fast",
      "display_name": "autojev/fast",
      "effective_context_window_percent": 95,
      "experimental_supported_tools": [],
      "input_modalities": [
        "text",
        "image"
      ],
      "priority": 0,
      "shell_type": "unified_exec",
      "slug": "autojev/fast",
      "support_verbosity": false,
      "supported_in_api": true,
      "supported_reasoning_levels": [],
      "supports_reasoning_effort_updates": false,
      "supports_reasoning_summaries": false,
      "supports_search_tool": false,
      "truncation_policy": {
        "limit": 10000,
        "mode": "tokens"
      },
      "use_responses_lite": false,
      "visibility": "list"
    }
  ]
}

Restart the agent after editing so it reloads its settings. Codex integration guide →

AutoJev writes this for you with one click and restores your original config

Connect an agent in AutoJev and it backs up the original file, writes these settings, and restores your original bytes when you disconnect or quit — keeping later non-conflicting edits. No hand-editing, no stale configs.

Frequently asked questions