Senlay Physical Context

Senlay — Physical-World Intelligence for Autonomous Systems

Senlay combines available environmental observations and forecasts into location-specific context for autonomous systems and AI applications, with sources, freshness, disagreements and missing information exposed through REST and MCP. The current product is an evaluation MVP.

This repository contains Senlay Open Network: the open station protocol, JavaScript SDK and edge-gateway starter. It is one component of Senlay, for connecting independently operated sensors.

senlay.cloud · API documentation

Connect an AI agent with MCP — no install

Add the public stateless Streamable HTTP endpoint to an MCP-capable client. Anonymous trial access requires no signup or OAuth:

{

"mcpServers": { "senlay": { "url": "https://senlay.cloud/mcp" } } }

The public MCP server exposes sense_location, check_operability, list_domains, and get_forecast_trust. Historical forecast trust can remain UNKNOWN when eligible archived pairs are insufficient; the demo history chart does not calculate a new trust score. It preserves Senlay evidence provenance and explicitly reports unavailable inputs instead of inventing values. To use an account tier, add Authorization: Bearer sl_live_... to the server headers. Operability results are decision support; humans remain the final authority.

Official MCP Registry metadata is provided in server.json. Version 1.0.0 is published as io.github.smartsurfsolar/senlay after live endpoint and schema validation.

What is open here

• vendor-neutral observation envelope and JSON Schema
• deterministic validation and normalization
• HMAC request signing for provider-to-gateway transport
• JavaScript client for any compatible ingestion endpoint
• small edge gateway that validates before forwarding
• adapter example and automated tests

The production aggregation engine, global registry, authentication, billing, provider credentials, historical place-memory corpus, risk models, and commercial connectors remain in the private senlay-platform-core repository.

The public website source is maintained separately in Senlay; its canonical domain is senlay.cloud. See repository boundaries for the public/private split and branch policy.

Status

This is the first 0.x protocol release. Self-service station registration and signed observation ingestion are available at senlay.cloud for authenticated Senlay accounts. The server retains a derived signing key that remains sensitive and can authenticate station submissions. Save the original station secret when returned and rotate it if lost or exposed.

The network is at the start of its path: there are no coverage guarantees and no SLA. It is not claimed as generally available.

Connect a station to the live network

First create a normal account at senlay.cloud/register.html, then create an API key in the dashboard and save it when shown. Stored keys cannot be retrieved; create a replacement if yours is lost. Use that account key for provider and station management; use the station secret only to sign observations.

# Create a provider with your account API key.

curl -sS -X POST https://senlay.cloud/api/v1/providers/register \ -H "Authorization: Bearer $API_KEY" -H 'content-type: application/json' \ --data '{"name":"My community network","slug":"my-community-network"}'

Substitute the returned prv_... ID. Save the returned station secret now.

curl -sS -X POST https://senlay.cloud/api/v1/stations/register \ -H "Authorization: Bearer $API_KEY" -H 'content-type: application/json' \ --data '{"providerId":"prv_REPLACE_ME","name":"Beach wind station","sensorType":"weather_station","location":{"lat":15.8801,"lng":108.3380}}'

Before publishing live data, use the dashboard's Send signed test action to verify the station secret and payload shape without storing a fake observation. Then publish to https://senlay.cloud/api/v1/observations using the returned stationId plus the X-Senlay-* signature headers described in the network protocol.

Nearby accepted observations appear in later authenticated /api/v1/sense and /api/v1/pwm responses as explicitly labelled direct station evidence. Checked station measurements also contribute to Senlay local memory baselines for the same place and season; those baselines are historical context, not current evidence.

Five-minute Python station

The Python connector uses only the standard library, so a Raspberry Pi or ordinary computer needs no package installation. Copy the stationId and one-time station secret returned by the dashboard, then preview a simulated observation locally:

git clone https://github.com/smartsurfsolar/senlay-platform.git

cd senlay-platform export SENLAY_STATION_ID='stn_REPLACE_ME' export SENLAY_STATION_SECRET='sl_station_REPLACE_ME' python3 python/simulator.py --seed 1

Preview mode does not contact Senlay. To verify the signature and payload without storing a reading, add the account API key and use the commissioning endpoint:

export SENLAY_ACCOUNT_API_KEY='sl_live_REPLACE_ME'

python3 python/simulator.py --dry-run --seed 1

To exercise live ingestion with generated data, use --publish. Simulator readings are always marked estimated, carry simulated: true, and never enter the checked-reading place-memory baseline.

python3 python/simulator.py --publish --count 3 --interval 10 --seed 1

For a real wind sensor, pass its readings to the publisher. The default quality is raw; use --quality checked only when your adapter and sensor calibration justify that claim.

python3 python/publish_reading.py \

--lat 15.8801 --lng 108.3380 \ --wind-speed 7.4 --wind-direction 62

Import SenlayClient and observation from python/senlay_client.py to connect another sensor process directly.

Try the live API — no install

CREDS="$(curl -sS -X POST https://senlay.cloud/api/v1/agent-register -H 'content-type: application/json' --data '{"ownerEmail":"you@example.com","agentName":"my-first-agent"}')"; printf '%s\n' "$CREDS"

API_KEY="$(printf '%s' "$CREDS" | sed -n 's/.*"apiKey":"\([^"]*\)".*/\1/p')" curl -sS -H "Authorization: Bearer $API_KEY" "https://senlay.cloud/api/v1/sense?lat=15.8801&lng=108.3380&field=kitesurfing"

Save both the apiKey and recoveryToken shown by the first command; each is returned only once. Replace you@example.com and my-first-agent with your own identity.

Local SDK and gateway

npm install

npm test cp .env.example .env npm run gateway

Send the example observation to the local gateway:

npm run example

The gateway validates observations locally. Set SENLAY_INGEST_URL to forward them to an authorized compatible ingestion endpoint.

MQTT bridge (optional)

For hardware that publishes JSON observations to an MQTT broker, configure the MQTT variables in .env and run:

npm run mqtt-bridge

The bridge subscribes to SENLAY_MQTT_TOPIC, validates every message through the SDK, signs it with the station secret, and forwards it to the HTTPS ingestion endpoint. Use one bridge credential per station; do not share a station secret among devices.

Branches

• main: stable public releases
• develop: active public development

Changes move from a feature branch to develop, then reach main after review and tests. See CONTRIBUTING.md.

Minimal SDK example

import { SenlayClient } from '@senlay/open-network';

const client = new SenlayClient({ endpoint: 'https://senlay.cloud/api/v1/observations', providerId: 'prv_REPLACE_ME', signingSecret: process.env.SENLAY_STATION_SECRET });

await client.publish({ stationId: 'stn_REPLACE_ME', observedAt: new Date().toISOString(), location: { lat: 15.8801, lon: 108.3380 }, measurements: [{ phenomenon: 'wind.speed', value: 7.4, unit: 'm/s' }] });

License

Apache-2.0. Contributions are welcome under the same license.