Integrations

What's live

Two inverter brands have full end-to-end integrations running on production sites today, plus one weather provider:

• SolarEdge — sites, equipment, telemetry, alarms. Polled every 15 minutes. The Layout and Sensors APIs need a higher SolarEdge monitoring tier; we degrade gracefully if you don't have it (we just don't render the per-module roof layout).
• Solis Cloud — sites, inverters, strings, telemetry, alarms. Polled every 5 minutes. Authenticate with the API ID + secret from your Solis console.
• Open-Meteo — hourly irradiance, temperature, cloud cover. Used to build the weather-normalised expected generation curve and to gate the alert detections.

What's in beta

Sigenergy has an adapter shipped that's awaiting live-site verification. Hybrid inverter / per-MPPT / battery topology is built; we're pairing the adapter with a real mySigen account before flipping it to "live". If you've got Sigenergy hardware you'd like to be the first live customer on, get in touch.

What's on the roadmap

Every other major inverter brand is on the roadmap, demand-driven. Sungrow, Fronius, SMA, Enphase, Huawei, GoodWe, Growatt, FIMER. New adaptors typically take around two weeks of work, and the first one is at cost for any customer who's making it possible.

If your portfolio runs a brand we don't have yet, tell us. We build to demand — a brand goes from "roadmap" to "live" when someone says they need it.

How to connect an integration

1. Settings → Integrations.
2. Click "Connect" next to the brand.
3. Paste the credentials — API key for SolarEdge, API ID + secret for Solis Cloud.
4. SolarFleet validates the credentials against the upstream API in real time. If they're wrong you'll know immediately.
5. Choose which sites to import. Defaults to "all sites the API exposes" but you can pick a subset.

Importing pulls equipment, recent telemetry (about two weeks by default), and any open alarms. Historical backfill runs in the background — typically three months of history is in place within an hour of connecting.

Credential handling

Credentials are encrypted with AES-256-GCM at rest, keyed off a worker secret that isn't accessible from any app code path. We never log credentials to disk in cleartext. The decrypted value lives in memory only for the duration of each upstream API call.

If you rotate a credential upstream (you should, regularly), update it in Settings → Integrations and the next polling tick uses the new value. There's no downtime — we'll fall back to the previous credential for the in-flight tick and switch cleanly.

Manual sites

You can add sites for brands we don't yet integrate with. They won't auto-import telemetry, but everything else works the same — cases, maintenance visits, BS EN 62446-1 reports, client PDFs. When we ship the integration for that brand, you can flip the existing manual site to "connected" without losing any of its history.