Open-Source Observability · zapdos-labs

unblink

Unblink is an AI-powered camera monitoring application that uses Vision Language Models (Qwen3-VL) to analyze live video feeds in real time. It runs as a private node in your network that connects to a relay server, enabling natural language search and chat interactions with your camera data while keeping video processing local.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/zapdos-labs/unblink

1.4k

GitHub stars

170

Forks

Primary language

AGPL-3.0

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	zapdos-labs/unblink
Owner	zapdos-labs
Primary language	Go
License	AGPL-3.0 — OSI-approved
Stars	1.4k
Forks	170
Open issues	3
Latest release	Unknown
Last updated	2026-03-09
Source	https://github.com/zapdos-labs/unblink

What unblink is

Go-based distributed architecture with a relay server and edge nodes that process RTSP/MJPEG streams using Qwen3-VL for frame analysis. Frontend built with SolidJS/TypeScript; backend uses PostgreSQL and go2rtc for protocol handling; communication via protobuf-defined services.

Quickstart

Get the unblink source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/zapdos-labs/unblink.git2cd unblink3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Private On-Premise Camera Networks

Organizations needing AI-powered video analysis without sending raw footage to cloud services. The node-relay architecture keeps processing local while maintaining central management.

Natural Language Video Search

Security/compliance teams wanting to query recorded frames using conversational language ('find footage of people in hard hats') rather than timestamp browsing or traditional filters.

Multi-Camera Monitoring with Unified Interface

Facilities with heterogeneous camera sources (RTSP, MJPEG) needing a single pane of glass with VLM-powered anomaly summarization across feeds.

Implementation considerations

Requires PostgreSQL backend and Go runtime (or pre-built binaries); plan infrastructure for relay server and edge node deployment.
VLM (Qwen3-VL) inference cost and latency depends on GPU availability; CPU-only inference will be slow. Budget for appropriate hardware.
Network architecture must accommodate relay-to-node communication; ensure firewall/NAT rules permit secure node registration and proxying.
RTSP/MJPEG camera compatibility must be validated before deployment; custom transport adapters may be needed for proprietary systems.
Environment variable configuration (`.env` pattern) required; no clear multi-tenant or role-based access control documentation visible.

When to avoid it — and what to weigh

No In-House DevOps Capacity — Requires running and maintaining a relay server plus edge nodes; not a turnkey SaaS. Needs operational oversight for uptime and configuration management.
Strict Commercial Licensing Requirements — AGPL-3.0 mandates source disclosure of any modifications and derived works served over network. Incompatible with proprietary commercial software strategies without legal review.
High-Frequency Real-Time Alerting — VLM inference adds latency. If sub-second threat detection is critical, traditional rule-based or lightweight model approaches are more suitable.
Unknown/Legacy Camera Protocols — Supports RTSP/MJPEG via go2rtc. Proprietary or obscure protocols will require custom integration work.

License & commercial use

AGPL-3.0 (GNU Affero General Public License v3.0). Copyleft license requiring source disclosure of modifications and derivative works, especially when served over a network as a web service.

AGPL-3.0 is not a permissive license. Commercial use is possible but carries legal obligations: any modifications or instances served to users must have source code available under AGPL-3.0. Using unmodified Unblink in a commercial product requires that the entire system either comply with AGPL-3.0 or be analyzed by counsel. Proprietary extensions or SaaS offerings require careful license review. Redistribution of derivatives demands source availability.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Limited
License clarity	Clear
Deployment complexity	High
DEV.co fit	Good
Assessment confidence	Medium

Security considerations

Node-relay model reduces cloud egress for video frames, keeping processing local. However, verify relay server security posture (authentication, TLS, rate-limiting not explicitly documented). AGPL-3.0 source requirement provides some transparency. No mention of frame encryption at rest/in-transit, access control granularity, or audit logging. Validate camera credential handling and network isolation before deployment. RTSP/MJPEG sources may carry known protocol vulnerabilities; go2rtc's security posture requires independent review.

Alternatives to consider

Frigate

Open-source NVR with object detection (YOLO); better documentation and more mature ecosystem. Permissive license (MIT). Lighter VLM integration; primarily suited for Frigate-native deployments.

Deepstack / CodeProject.AI

Edge AI platform for video/image analysis with REST API and UI. Focuses on object detection and custom models. Simpler single-node deployment than relay architecture.

Managed services with strong VLM/video analysis. No self-hosting complexity. Trade-off: data sent to cloud, ongoing per-frame costs, vendor lock-in.

Software development agency

Build on unblink with DEV.co software developers

Unblink combines edge VLM inference with distributed node-relay architecture. Ideal for privacy-first organizations. Requires Go/Postgres infrastructure and AGPL-3.0 license compliance review.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

notifiarr WatchAlert kubetools

Related on DEV.co

Explore the category and the services that help you build with it.

Open-Source Observability AI Application Development Cloud Deployment & Infrastructure API & Backend Development Talk to DEV.co

unblink FAQ

Can I use Unblink commercially without open-sourcing my modifications?

Not without legal review. AGPL-3.0 requires source disclosure for any modified version served to users. Using unmodified Unblink is permissible, but any derived or extended system must comply with AGPL-3.0 or undergo license analysis.

What VLM models are supported?

README specifies Qwen3-VL for frame analysis. No documentation on model swapping or alternative VLMs; may require code modification.

How do I run this in production?

README covers local development. Production guide for relay server, node clustering, load-balancing, and HA is not documented; requires custom engineering.

What inference hardware do I need?

Unknown. VLM performance depends on GPU; CPU-only feasibility not documented. Budget planning requires benchmarking with your hardware.

Custom software development services

Adopting unblink is usually one piece of a larger software development effort. As a software development agency, DEV.co provides software development services and web development expertise — pairing senior software developers and web developers with your team to design, build, and operate open-source observability software in production.

Ready to Deploy Private AI Video Monitoring?

Unblink combines edge VLM inference with distributed node-relay architecture. Ideal for privacy-first organizations. Requires Go/Postgres infrastructure and AGPL-3.0 license compliance review.

Talk to DEV.co Schedule a Consultation