Analysis updated 2026-08-13
Expose Kafka and MQTT streams to web and mobile apps over HTTP or WebSocket.
Give an AI agent one unified MCP endpoint for multiple servers, APIs, and Kafka.
Bridge IoT devices using MQTT to a Kafka-based system of record.
Apply authentication, schema validation, and observability at the gateway layer.
| aklivity/zilla | getactivity/shapeview | elder-plinius/v3sp3r | |
|---|---|---|---|
| Stars | 1,046 | 1,054 | 1,013 |
| Language | Java | Java | Java |
| Last pushed | — | 2026-05-01 | — |
| Maintenance | — | Maintained | — |
| Setup difficulty | moderate | easy | hard |
| Complexity | 4/5 | 2/5 | 4/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires Docker Compose, ready-made example configs are included for both gateway paths.
Zilla is a gateway server that sits between different types of applications, devices, and AI systems and helps them talk to each other and to a message streaming system called Apache Kafka, even when they do not natively speak the same protocol. For example, a web browser cannot talk to Kafka directly, and an AI agent might need to reach several different services, each with its own way of connecting, credentials, and data format. Zilla translates between these different protocols, including HTTP, MQTT, gRPC, and a newer protocol called MCP that AI agents use to discover and call tools. The tool provides two main capabilities. The first, called the Event Gateway, exposes Kafka and MQTT data streams to regular applications and devices using more familiar protocols like HTTP or WebSocket. The second, called the MCP Gateway, gives an AI agent one single, unified connection point that can reach multiple underlying MCP servers, existing HTTP APIs, and Kafka clusters, all organized under one catalog of available tools and capabilities. Both gateways are configured through a single settings file and share the same routing, security, and monitoring systems underneath. Zilla also handles authentication and authorization at the gateway level, meaning it can check who or what is making a request before forwarding it, validate and transform data formats like JSON, Avro, and Protobuf, and record detailed metrics about requests passing through it. It is designed to run efficiently with minimal added delay, keeping most of its internal processing stateless so it can scale across multiple instances. Getting started requires Docker Compose, and the project includes ready to run example setups for both the Event Gateway and MCP Gateway paths. It can also be installed directly through Docker or deployed to Kubernetes using a Helm chart. Zilla is released under the Aklivity Community License, a custom license with its own specific terms rather than a standard open-source license, and a separate paid edition called Zilla Plus adds enterprise features. This project is aimed at developers and platform engineers building event-driven systems or connecting AI agents to existing APIs and data streams.
A multi-protocol gateway that connects apps, devices, and AI agents to Apache Kafka and other services through HTTP, MQTT, gRPC, and MCP.
Mainly Java. The stack also includes Java, Kafka, Docker.
Released under a custom Aklivity Community License rather than a standard open-source license, review the terms before commercial redistribution.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.