| Service | URL | Purpose |
|---|---|---|
| PTX MCP Server | http://localhost:8001 | MCP server (tools at /mcp) |
| Open WebUI | http://localhost:3001 | Web UI for AI + MCP |
docker compose up -d # Start
docker compose ps # Status
docker compose logs -f # Logs (all)
docker compose down # Stop- Open http://localhost:3001 and create an account.
- In Admin Panel → Settings → Connections, add an OpenAI-compatible API (e.g. Ollama, OpenAI).
- Chat; the MCP tools are available to the model. Example prompts: "Get facts from the PTX chassis", "Show hardware inventory", "Get current configuration in text format".
Cursor / Claude Desktop — Add to MCP config:
{
"mcpServers": {
"ptx-mcp": {
"url": "http://localhost:8001/mcp",
"transport": "http"
}
}
}For remote access, use http://<your-server-ip>:8001/mcp. See Remote Access.
Copy .env.example to .env and set:
PTX_HOST— PTX hostname or IPPTX_USER— Login usernamePTX_PASSWORD— Login passwordPTX_PORT— Optional; default 22PTX_SSH_KEY— Optional; path to SSH private key (preferred over password)
Test PTX connection:
docker exec ptx-mcp-server python -c "
from tools.common import run_cli_command_on_ptx
ok, out = run_cli_command_on_ptx('show version')
print('OK:', out[:80] if ok else out)
"Container restarts — Check docker compose logs. Ensure .env has valid PTX_HOST, PTX_USER, PTX_PASSWORD (or PTX_SSH_KEY).
Port in use — Change host port in docker-compose.yml (e.g. "8002:8000").
This setup is for lab/development. For production: use SSH keys, HTTPS, and restrict network access. Do not commit .env.