In your Zephyr application, initialize the Golioth client and connect using PSK credentials; confirm the device appears online in the Golioth console
Use the Golioth LightDB Stream SDK API (golioth_stream_push_cbor or golioth_stream_push_json) to publish timestamped sensor readings to a stream path such as /sensors; call this from a periodic work queue item
In the Golioth console, navigate to Pipeline (Output Streams) and create a new pipeline; select the LightDB Stream data source for your project
Configure the pipeline destination as an HTTP webhook, specifying your backend endpoint URL and any required authentication headers
Trigger a few device readings and verify the webhook receives JSON payloads containing the device ID, timestamp, and sensor data fields
In your visualization tool (e.g., Grafana), connect to the webhook destination database or use the Golioth REST API to query historical stream data for charting
Known gotchas
LightDB Stream data is append-only and time-series; it cannot be updated or deleted after ingestion — design your data schema carefully before deploying to production
CBOR encoding is more compact than JSON and recommended for constrained devices; ensure your backend webhook handler can decode CBOR if you use that encoding
Pipeline webhook delivery is best-effort; if your endpoint is unavailable, events may be dropped depending on pipeline configuration — implement idempotent ingestion with a device ID and timestamp as the deduplication key
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp