FAQ

FAQ

Questions people evaluating Device Agent actually ask, before they've decided to adopt it. If you're already building with it, the tutorial guides are a better starting point.

Licensing & cost

Is it really free? Yes. The daemon, dashboard, and everything in this repository are Apache-2.0 — free to use, modify, and run for personal, lab, and commercial production use at no charge, subject to preserving notices (see NOTICE). See the README's License section.

What does "Enterprise" mean, then? Production support, SLAs, and Zyvor's other commercial products are licensed separately from this open-source agent. Contact sales@zyvor.dev or see zyvor.dev if you need that; nothing in this repository requires it.

Support

What if I find a bug? Open a GitHub issue on this repository.

What if I find a security vulnerability? Don't open a public issue — see SECURITY.md for private reporting instructions and the current v0.1 security boundary list.

Is there a support SLA? Not for the open-source project itself — community support is via GitHub issues. Contact sales@zyvor.dev for commercial support arrangements.

Production readiness

Is this production-ready? Status is v0.1.6 production-hardening — not GA, and not Minewing-silicon-qualified. The Linux agent (packages, auth/TLS, emulator/hil-ci-emulator, lab-surrogate) is shippable for hardened installs; physical Minewing HIL remains unsigned (minewing_claimable=false on surrogate evidence). Run make qualify and sign the hardware checklist only after physical HIL. See PRODUCTION.md, HIL.md, and TEST-REPORT.md. ROADMAP.md lists what is still ahead (e.g. local AI inference remains unscoped).

How stable is the API? Each REST endpoint and config field added is documented in the README under its introducing version (e.g. "v0.1.4"). No formal API-stability guarantee is made pre-1.0 — check CHANGELOG.md when upgrading.

Does Nodra MQTT need TLS? Prefer [nodra.tls] enabled = true (MQTTS) whenever the path leaves a trusted LAN. Plain MQTT is loopback/LAN-only.

Cloud & data

Does this require a cloud account or internet connection to run? No. Device Agent runs entirely on the device — inventory, health, REST API, dashboard and Prometheus metrics all work with zero network dependency beyond serving those locally. Publishing to Nodra (industrial protocol routing) and projecting into Fleet (remote lifecycle) are both optional integrations, off unless configured.

What happens if the network drops? The documented acceptance criterion (README's "First demo acceptance test") is explicit: the agent must "keep working during WAN loss" and "sync after reconnect through Nodra WAL" — WAN connectivity loss doesn't stop local hardware monitoring or health reporting; only upstream publishing queues until reconnect.

Where does my data go? Nowhere, by default. Nothing leaves the device unless you configure Nodra MQTT publishing or Fleet's inventory bridge — both point at endpoints you control, not a Zyvor-operated cloud.

Security

What's the security model? See SECURITY.md in full. In short: the daemon is intended for a trusted Linux edge node, plugin manifests must be administrator-provisioned (no arbitrary remote code execution), and the REST API supports bearer-token auth, mTLS (with optional TPM2-backed key storage), and independent transport TLS — see the README's "API auth and Unix socket", "mTLS enrollment", and "TLS" sections.

Does it operate its own certificate authority? No — enroll is a client submitting a CSR to an enrollment endpoint you run; key custody, revocation and rotation are your organization's responsibility. See docs/MTLS_ENROLLMENT.md.

Extensibility

Can I add my own sensors? Yes — the sensor plugin protocol (docs/PLUGIN_PROTOCOL.md) is a documented contract: an executable plus a JSON manifest, run out-of-process, with optional privilege-drop and seccomp sandboxing.

Can I add my own camera/video processing? Device Agent captures and serves live snapshot/MJPEG streams (docs/CAMERA.md); it deliberately does not run inference on frames itself — that's future, still-unscoped work per docs/ROADMAP.md's "Edge AI bridge" entry.

Hardware

What hardware is supported? See the README's "Is this for you?" section and docs/REFERENCE_HARDWARE.md — a capability profile (required buses + recommended buses), not a fixed list of certified boards.

Does it manage a fleet of devices? No — Device Agent runs one instance per device and projects its inventory/health into Zyvor Fleet, which owns fleet-wide lifecycle and desired-state management. Device Agent itself has no multi-device concept.

Platform

What OS/architecture does it run on? Linux only. arm64 is first-class; amd64 works for development, CI, and containers. No Windows or macOS support (macOS can build and run the daemon for development, but hardware discovery returns mostly empty results there — see 1. Getting started).