zenfeed

zenfeed is an MCP server that turns RSS feeds into an AI-assisted information hub. It can act as a server in front of an RSS source and expose a Query API, an RSS export API, and a Notify webhook, all configurable through YAML-based rewrite and routing rules. The README notes that zenfeed supports features like daily briefings, monitoring, and custom data sources, and it can be integrated with Cherry Studio MCP or used as an MCP endpoint for RSSHub. To get started, deploy the service with Docker and supply your API key, then connect your RSS sources and configure MCP-based routing and notifications as needed. When exploring capabilities, you can configure and test: the Query API for programmatic access to processed feed data, the RSS Exported API for feeding external RSS clients, and the Notify Webhook to push alerts or summaries to other systems. The web UI (via the provided docker-compose setup) allows you to add sources, configure daily briefs, and tune processing through a set of declarative YAML configurations.

Prerequisites:

• Docker and Docker Compose installed on the host
• Basic familiarity with YAML configuration

Installation steps:

1. Prepare environment and prerequisites

• Ensure you have Docker installed: https://docs.docker.com/get-dstarted/
• Have a valid API key ready for zenfeed usage

2. Retrieve configuration

• Download the docker-compose.yml from the zenfeed repository (or reference the README example):

# Mac / Linux: download the configuration file
curl -L -O https://raw.githubusercontent.com/glidea/zenfeed/main/docker-compose.yml

3. Start the service

• Start the service in detached mode, providing your API key:

API_KEY="sk-..." docker-compose -p zenfeed up -d

4. Verify and access

• Visit http://localhost:1400 to access zenfeed UI and configure sources, daily briefs, and MCP routing as needed.

Notes:

• If deploying on a VPS or public network, ensure firewall rules allow 1400 and consider securing the API key access.
• zenfeed currently does not include an authentication mechanism; restrict access to trusted IPs as appropriate.

Tips and notes:

• The default setup relies on Silicon Flow model services. Models include Qwen/Qwen3-8B (free) and Qwen/Qwen3-Embedding-4B. If you want to use other providers or customize models, edit the docker-compose.yml and corresponding config docs in docs/config-zh.md.
• The REST-like MCP configuration is YAML-based; you can customize pipelines, prompts, and routing rules via the provided docs (Rewrite Rules, config-zh.md).
• When running publicly, be mindful of API_KEY exposure since zenfeed initially lacks authentication; use strict firewall rules and consider adding a reverse proxy with authentication if needed.
• To connect zenfeed to an MCP client like Cherry Studio, use the default MCP URL http://localhost:1301/sse or adjust to your deployment topology.