433 / 868 MHz · passive reception

The map of the ambient noise your connected devices already broadcast in the clear.

Weather stations, pressure sensors, water pumps — thousands of home devices announce their state in the clear on the ISM band. IQLOG maps them, with the community.

Frames captured
Active stations
Distinct models
Families covered
/ 9

Detections map

precise coordinates · full history

Sensor families

anonymized aggregation

Latest frames log

Join the collection

01

Plug in an RTL-SDR dongle

Any RTL-SDR dongle (~€25) plugged into your PC, homelab, or Android phone over USB OTG is enough to start listening.

02

Run the agent

The Python agent or the Android app run in the background, decode with rtl_433, and only send unencrypted frames.

03

See your contribution

Each install gets its own API key. Your frames join the map, anonymized before any public display.

Linux · Raspberry Pi · homelab Install the agent in one line
curl -fsSL https://iqlog.net/install.sh | sh

Installs rtl_433 (latest, compiled), sets up a systemd service that captures 24/7 and registers itself, then add it under “My devices”. Also available: Windows (.exe) and Android apps.

// connect an agent or the app frame format identical to rtl_433 -F json
API URL
Ingestion (key auth)
POST /api/ingest
Validate a key
GET /api/keys/validate

Key authentication: Authorization: Bearer <key> header. Keys are issued by the instance administrator, one per install.

// talk to IQLOG from any AI · MCP read-only · no key · Streamable HTTP

Connect any MCP-capable assistant (Claude & co.) to the endpoint below — it can then search and aggregate the live captures on its own.

https://iqlog.net/mcp

In Claude: Settings → Connectors → Add custom connector, then paste the URL. Tools: search, stats, object_frames, facets, opensearch_query.

Ask an AI, for example:
  • “Which weather stations are near 50.63, 3.05, and what's their latest temperature?”
  • “Break down the last 24 h of frames by category, per hour.”
  • “List every tyre-pressure (TPMS) model seen this week and its distinct sensor count.”

Machine-readable discovery for AI clients: /.well-known/mcp and /llms.txt.