MCPheonix

MCPheonix is a lightweight, self‑healing Model Context Protocol (MCP) server built on Elixir’s Phoenix framework. It bridges AI assistants and application data by exposing a unified MCP interface that handles resource querying, tool invocation, and event streaming—all while running in a resilient edge‑first environment.

The server solves the common pain point of integrating AI models into production systems: developers need a way to expose application state, execute business logic, and stream updates without wrestling with low‑level networking or state consistency. MCPheonix abstracts these concerns behind a JSON‑RPC endpoint for synchronous requests and a Server‑Sent Events stream for real‑time notifications. By combining Phoenix’s robust web stack with Cloudflare Durable Objects and Workers, the server achieves automatic recovery—when a node fails, its state is restored from durable storage and re‑synchronised across the edge network. This guarantees continuous operation even during regional outages or scaling events.

Key capabilities include:

• Simple resource system: CRUD operations on application data via MCP resources, allowing AI assistants to read and mutate state directly.
• Event publish/subscribe: Tools can emit events that are streamed to subscribed clients, enabling reactive workflows and real‑time dashboards.
• Tool invocation: The server can call out to external services (e.g., Flux for image generation or Dart for task management) and return results through MCP tools, giving AI models access to specialized functionality without exposing raw APIs.
• Edge distribution: All critical state is replicated across Cloudflare’s global edge network, ensuring low latency and high availability for geographically dispersed users.
• Extensibility: A JSON‑based configuration lets developers plug in custom MCP servers, commands, or environment variables, making the platform adaptable to a wide range of use cases.

In practice, MCPheonix shines in scenarios where AI assistants need to orchestrate complex workflows: a conversational agent that schedules meetings, generates visual content on demand, or monitors system health in real time. By handling the plumbing—state persistence, fault tolerance, and event delivery—the server lets developers focus on business logic while AI models interact seamlessly through a single protocol.