Open-source Tesla intelligence on your infrastructure: recorded driving and charging history, live vehicle views, analytics, alerts, automations, and optional Helix AI. Start with Getting Started, not the AI setup.
The software is MIT-licensed. Budget for hosting, storage, backups, a domain, Tesla API/telemetry usage under Tesla's current billing terms, and any external integrations. Hosted AI providers may charge separately. Local Ollama avoids hosted-model fees, not hardware and electricity costs.
Docker with Compose v2 is the installation path. Resource needs depend on vehicle count, signal frequency, retention, enabled services, and local AI models. There is no universal sizing guarantee; monitor disk, memory, CPU, and database growth on your workload. Go and Node.js are only needed for source development.
Availability depends on Tesla's Fleet API access, region, vehicle hardware, firmware, owner permissions, and configured signals. Do not assume every model exposes every field or command. Tesla's Fleet Telemetry compatibility notes identify firmware requirements and older Model S/X exclusions. Check your vehicle's reported configuration and errors during setup.
No. OAuth grants account access; it does not deploy a receiver or subscribe a
vehicle. Follow Fleet API setup, then
Fleet Telemetry. Use Settings → Fleet Setup
(/settings/fleet-setup) to select the vehicle, signals, and intervals and submit
the configuration. Wait for synchronization and verify fresh incoming data.
No. Tesla's official telemetry setup includes pairing the application's virtual key and configuring the signing proxy before sending the telemetry configuration. Application registration, pairing, and a reachable mTLS receiver are separate prerequisites.
Six: openid, offline_access, vehicle_device_data, vehicle_location,
vehicle_cmds, and vehicle_charging_cmds. Missing grants affect the corresponding
features. See the scope reference
and reauthorize after changing access.
Not safely without additional configuration. Open mode does not authenticate users, and Compose publishes several ports. Use a trusted authenticating proxy, overwrite the configured identity header, block direct API access, configure TLS and strong secrets, and test backups. See the production checklist. The receiver's public mTLS endpoint and anonymous public-key URL need separate routing.
Compose uses .env for interpolation, not automatic forwarding. A variable must
be mapped into the service environment. See
required overrides, especially
for ENCRYPTION_KEY and TESLA_COMMAND_PROXY_URL.
No. Tesla connectivity uses Tesla services, and enabled integrations can send data externally. Helix is optional; hosted providers receive request context. Review Helix privacy controls, configured destinations, and provider terms. Redaction reduces exposure but is not a guarantee that no sensitive information leaves the deployment.
Retention depends on the deployed schema and retention jobs. Storage is not unlimited, and aggregate data is not a replacement for raw history in every view. Check your deployment's policies and disk growth, then establish and test backups. Historical data before installation is not automatically reconstructed.
Check vehicle connectivity and configuration synchronization, then the Fleet Setup error panel, receiver logs, MQTT, API ingest, and browser connectivity. A sleeping vehicle or unchanged signal need not publish continuously. UI fallback polling is not proof that new vehicle telemetry is arriving. Continue with Troubleshooting.
Redis provides shared live reads and fanout, but does not by itself make vehicle state machines and reconciliation active-active. Plan vehicle ownership, leases, or affinity before scaling ingest processes. See Architecture and Kubernetes deployment.
Use the contributor entry guide for your first change. For an issue, include the version, deployment method, reproduction steps, and redacted diagnostics. Never post tokens, private keys, VINs, or precise locations. Report vulnerabilities through the repository's security policy.