luneski/mqtt-explorer
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add observability for LLM topic context inclusion (#1038)

Browse Source 
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: thomasnordquist <7721625+thomasnordquist@users.noreply.github.com>
Co-authored-by: Thomas Nordquist <thomasnordquist@users.noreply.github.com>
This commit is contained in:
Copilot
2026-01-30 20:53:29 +01:00
committed by GitHub
parent 080a773dbd
commit ed8a7f559e
194 changed files with 35234 additions and 4085 deletions
Download Patch File Download Diff File Expand all files Collapse all files

33
DOCKER.md
View File

@@ -71,6 +71,39 @@ docker-compose up -d
| `UPGRADE_INSECURE_REQUESTS` | No | `false` | Set to `true` to enable CSP upgrade-insecure-requests directive. **Only use when deployed behind an HTTPS reverse proxy (nginx, Traefik, etc.) with valid SSL certificates.** This upgrades all HTTP requests to HTTPS and will break direct HTTP access. |
| `X_FRAME_OPTIONS` | No | `false` | Set to `true` to enable X-Frame-Options: SAMEORIGIN header to prevent clickjacking. **Disables iframe embedding when enabled.** |
### AI Assistant / LLM Configuration
| Variable | Required | Default | Description |
|----------|----------|---------|-------------|
| `LLM_PROVIDER` | No | `openai` | AI provider to use (`openai` or `gemini`) |
| `OPENAI_API_KEY` | No | - | OpenAI API key for AI Assistant (provider-specific) |
| `GEMINI_API_KEY` | No | - | Google Gemini API key for AI Assistant (provider-specific) |
| `LLM_API_KEY` | No | - | Generic API key for AI Assistant (works with either provider) |
| `LLM_NEIGHBORING_TOPICS_TOKEN_LIMIT` | No | `500` | Token limit for neighboring topics context in AI queries (increased for better device relationship detection) |
**Architecture**: The backend proxies all LLM API requests via WebSocket RPC. API keys are **never** sent to the frontend - only an availability flag is transmitted. The frontend calls the backend via WebSocket RPC (`llm/chat` event), and the backend makes requests to OpenAI/Gemini on behalf of the client.
**Security**:
- ✅ API keys remain server-side only
- ✅ Keys never embedded in client bundles
- ✅ Keys never transmitted to frontend
- ✅ Backend controls all LLM access
- ✅ Communication via secure WebSocket RPC
**Note**: If no LLM environment variables are set, the AI Assistant feature will be completely hidden from users.
**Example with AI Assistant**:
```bash
docker run -d \
-p 3000:3000 \
-e MQTT_EXPLORER_USERNAME=admin \
-e MQTT_EXPLORER_PASSWORD=secret \
-e LLM_PROVIDER=openai \
-e OPENAI_API_KEY=sk-proj-xxxxxxxxxxxxxxxxxxxx \
-e LLM_NEIGHBORING_TOPICS_TOKEN_LIMIT=500 \
ghcr.io/thomasnordquist/mqtt-explorer:latest
```
### Authentication Modes
**Standard Mode (Default):**