dunglas/mercure

Health v2 · maintenance only

Adoption not part of the score

5300 stars · 346 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Mercure is an open protocol (with an IETF Internet-Draft) for pushing real-time updates to browsers and HTTP clients over Server-Sent Events, plus a production-grade reference hub written in Go. It serves as a fast, battery-efficient alternative to WebSockets, with JWT-based authorization, topic selectors, and Last-Event-ID replay.

Use cases

• push real-time updates from a web API to browsers and mobile apps
• stream LLM tokens and AI agent progress to users
• add GraphQL subscriptions or async capabilities to REST APIs
• build live dashboards and notifications
• implement collaborative editing and Hotwire/Turbo Streams page updates
• broadcast events to IoT devices over plain HTTP

When to choose

• you need server-to-client push over HTTP/SSE without client-side SDKs
• you want automatic reconnection, replay, and JWT-secured private updates
• you need a self-hostable or managed real-time layer with Docker/Kubernetes support
• you're streaming LLM responses or agent steps like ChatGPT-style APIs

When to avoid

• you need bidirectional, client-to-server streaming (use WebSockets)
• you require multi-node clustering on the free community edition (clustering transports are paid/Enterprise)
• your project cannot comply with the AGPL-3.0 license

Facets

service · maturity stable

http-server websocket middleware auth monitoring api-framework web-development apis backend large-language-models windows self-hosted cloud go server-sent-events sse mercure realtime push-updates event-streaming llm-streaming hypermedia caddy jwt-authorization real-time linux macos docker kubernetes

Member repositories

For agents

markdown · JSON · MCP: product_card(name="dunglas/mercure")