Nexus Mythos (nexusai)

CLI agent based on the Claude Code-style workflow, adapted to:

• Use local or hosted OpenAI-compatible APIs (OpenRouter, Ollama, LiteLLM, etc.)
• Provide an OpenClaw-style Live Screen viewer for GUI automation workflows
• Keep config under ~/.nexusmythos/ (Windows: C:\Users\<user>\.nexusmythos\)

Install (global)

After publishing to npm under the configured name:

npm install -g @henrq-crdso/nexus-mythos

Run:

nexusai

Settings file

Create:

• Windows: C:\Users\<user>\.nexusmythos\settings.json
• macOS/Linux: ~/.nexusmythos/settings.json

Example (OpenRouter):

{
    "env": {
        "ANTHROPIC_BASE_URL": "https://openrouter.ai/api/v1",
        "ANTHROPIC_API_KEY": "sk-or-v1-...",
        "ANTHROPIC_MODEL": "qwen/qwen3.6-plus:free" // example 
    }
}

Example (Ollama local):

{
  "env": {
    "CLAUDE_CODE_USE_OPENAI": "1",
    "OPENAI_BASE_URL": "http://localhost:11434/v1",
    "OPENAI_MODEL": "llama3.1"
  }
}

Live Screen (OpenClaw-style)

Start the CLI and a lightweight local viewer:

nexusai live

This starts a local HTTP server (default http://localhost:7331) and prints the URL. The GUI automation tool writes the latest screenshot to:

~/.nexusmythos/live/last.png

Enable GUI automation tools (mouse/keyboard + screenshots) by setting:

{
  "env": {
    "NEXUS_ENABLE_COMPUTER": "1"
  }
}

Optional dependencies

GUI automation uses optional deps:

• screenshot-desktop
• @nut-tree/nut-js

If they fail to install on a platform, the CLI still runs; only the Computer tool is disabled.