AutoJev
Download

Integration · Agent

Use any model in Hermes with one local gateway

Hermes is Nous Research’s agent, configured through ~/.hermes/config.yaml. AutoJev adds itself as a Chat Completions provider there, so Hermes can use any model or route you set up — and every provider key stays in one place.

  • One click edits ~/.hermes/config.yaml
  • Chat Completions adapter
  • All selected models and routes listed
  • Original YAML restored on disconnect
Download AutoJevStar on GitHub

Agents

+ Agent

CodexResponses

~/.codex/config.toml

Connected

Claude CodeMessages

~/.claude/settings.json

Connected

Gemini CLIGemini

~/.gemini/settings.json

Connected

OpenCodeChat

~/.config/opencode/opencode.json

Connect

How AutoJev connects Hermes

Hermes reads its model settings from a YAML file. When you connect it, AutoJev sets the model block to the autojev provider at http://127.0.0.1:9527/v1 with api_mode chat_completions, and adds providers.autojev with the list of models and routes you selected.

Hermes therefore always speaks Chat Completions to the gateway. Behind it, AutoJev picks the upstream model, fails over when a provider errors before output starts, and converts to Responses or Anthropic Messages if the chosen model needs it. Hermes never sees the difference.

What AutoJev configures

ItemValue
Config file~/.hermes/config.yaml (HERMES_HOME is respected)
AdapterChat Completions only
Model blockmodel.provider = autojev, model.default = your default
Providerproviders.autojev with base URL, models and default_model
Keyautojev-local-hermes (placeholder; labels requests as hermes)
Headerextra_headers: x-autojev-agent = hermes

What Hermes gains from the gateway

One place for keys

Provider keys live in AutoJev on your machine. Hermes only holds a placeholder key, and requests go directly to your providers with no markup.

Failover before output

On connection errors or HTTP 401/402/403/408/429/5xx before output starts, a route retries another candidate (up to 4 attempts by default) and respects Retry-After.

Per-agent logs and cost

Every request is labeled hermes, with model, route, input/output/cache tokens, latency and estimated cost. Prompts and responses are not stored.

Routing without edits

Connect a route once. Switch its strategy between load balancing and intelligent selection in AutoJev without touching Hermes’s config again.

Set up Hermes with AutoJev

  1. 1

    Install AutoJev

    Download the desktop app and open it. The gateway runs at http://127.0.0.1:9527.

  2. 2

    Add providers and models

    Add DeepSeek, OpenRouter, OpenAI, Ollama or any OpenAI-compatible or Anthropic-compatible endpoint with your own key, test it and add models.

  3. 3

    Optional: create a route

    A route such as autojev/fast lets AutoJev choose between models with a load-balancing or intelligent-selection strategy.

  4. 4

    Connect Hermes

    On Agents, AutoJev detects the hermes executable. Select models and routes, choose the default, and click Connect.

  5. 5

    Run Hermes

    Start Hermes as usual; its requests appear in AutoJev Logs under hermes.

    bash
    hermes

The config AutoJev writes

Example with the route autojev/fast as default plus one single model. Other keys in your YAML are kept.

~/.hermes/config.yaml
# ~/.hermes/config.yaml — keys AutoJev writes on connect model: provider: autojev default: autojev/fast base_url: http://127.0.0.1:9527/v1 api_key: autojev-local-hermes api_mode: chat_completions providers: autojev: name: AutoJev Model Gateway api: http://127.0.0.1:9527/v1 api_key: autojev-local-hermes transport: chat_completions default_model: autojev/fast models: autojev/fast: name: Fast openrouter/qwen/qwen3-coder-flash: name: Qwen 3 Coder Flash extra_headers: x-autojev-agent: hermes autojev_default_binding: fast

Switching models and routes

Every model and route you selected is listed under providers.autojev.models. To change the default or the list, edit the selection on the Agents page and reconnect, or run “Batch update configurations”. Restart Hermes to pick up the change.

If Hermes is connected to a route, you can change models without touching Hermes at all: edit the route’s strategy or candidates in AutoJev.

Restoring config.yaml

AutoJev validates the YAML before writing and keeps a backup of the original. Disconnect, “Stop and restore configurations” or quit reverts the keys it set while keeping non-conflicting edits you made later. YAML comments may be reformatted while connected.

Point Hermes at a route, not a model

Routes decide per request: weighted load balancing, or intelligent selection by cost, quality or speed. To pin one model, call <provider>/<model_id> directly.

Routes

Create reusable routes and bind your agents to a route ID.

+ Add route

fast

autojev/fast

quality

autojev/quality

cheap

autojev/cheap

balanced

autojev/balanced

lb

autojev/lb

Frequently asked questions

Which protocol does Hermes use with AutoJev?

Chat Completions. The Hermes adapter requires it, and AutoJev converts to whatever protocol the upstream model uses.

Can Hermes use DeepSeek or OpenRouter models?

Yes. Add the provider in AutoJev, select its models for Hermes on the Agents page, and connect.

Can Hermes use a local model?

Yes. Add the Ollama preset (http://127.0.0.1:11434) or any local OpenAI-compatible server, add a model, and select it for Hermes.

Does AutoJev respect HERMES_HOME?

Yes. If HERMES_HOME is set, AutoJev edits config.yaml in that directory instead of ~/.hermes.

What happens if I quit AutoJev?

Quitting from the tray restores Hermes’s configuration. On next launch AutoJev reconnects it, unless you disconnected Hermes manually.

Keep exploring

FeaturesOne-click agent setupConnect Codex, Claude Code, Gemini CLI, OpenCode and more. AutoJev edits each native config, backs it up and restores it.FeaturesSmart model routingExpose routes like autojev/fast to every agent. Load-balance models or let Jev pick by cost, quality, speed or balance.FeaturesFailover & load balancingWeighted pools, priority groups and sticky sessions, with automatic failover on 429s and 5xx errors and Retry-After cooldowns.IntegrationsOllamaUse local Ollama models in any coding agent, prefer them in routing, and fail over to cloud models when needed.Use casesFor multi-agent developersCodex, Claude Code, Gemini CLI, OpenCode and more on one local gateway, with per-agent models and one request log.

Updated 2026-10-14

Give every agent the right model.

Install AutoJev, add the providers you already pay for, and connect your agents in one click. Everything stays on your machine.

Download AutoJevStar on GitHub

Free & open source · macOS · Windows · Linux