freshrss-mcp
Enables AI assistants to read and manage FreshRSS feeds, including listing subscriptions, browsing articles, marking read/unread, and starring.
README
freshrss-mcp
An MCP server that puts a small, well-typed toolset in front of a FreshRSS instance, so an AI assistant (Claude, etc.) can read and manage your feeds: list subscriptions, browse unread/starred articles, read full content, mark read/unread, star, add feeds, and mark whole streams read.
It talks to FreshRSS over its built-in Google Reader-compatible API
(/api/greader.php), so it works with any standard FreshRSS install — no
plugins or schema changes required.
The server speaks MCP over streamable HTTP on a single port and gates the
endpoint with a static bearer token. /healthz is open for liveness/readiness.
Tools
| Tool | Description |
|---|---|
whoami |
Authenticated user + instance URL — a quick connectivity/credentials check. |
list_feeds |
Subscribed feeds with category and unread count. |
list_categories |
Categories/folders with total unread count. |
list_articles |
Articles (newest first), filterable by unread/starred/all, by feed or category, with an optional contains substring filter and pagination. |
get_article |
Full HTML content of a single article by id. |
mark_read / mark_unread |
Toggle read state for one or more article ids. |
star / unstar |
Toggle the star/favorite for one or more article ids. |
mark_all_read |
Mark an entire stream (all, a feed, or a category) read, optionally only items older than N seconds. |
add_feed |
Subscribe to a new feed URL, optionally filed under a category. |
Article ids returned by list_articles/get_article are the values you pass
back to the mutating tools.
Configuration
All configuration is via environment variables (see .env.example):
| Variable | Required | Description |
|---|---|---|
FRESHRSS_URL |
yes | Base URL of your FreshRSS instance, without /api/... (e.g. https://rss.example.com). |
FRESHRSS_USER |
yes | FreshRSS username. |
FRESHRSS_API_PASSWORD |
yes | The per-user API password (see below) — not the web login password. |
MCP_TOKEN |
recommended | Bearer token clients must send. If unset, the MCP endpoint is unauthenticated — only acceptable for purely local use. |
PORT |
no | Listen port (default 8080). |
FRESHRSS_TIMEOUT |
no | Per-request timeout to FreshRSS, seconds (default 30). |
Enabling the FreshRSS API
- In FreshRSS, go to Settings → Authentication and enable "Allow API access" (the global toggle for the GReader/Fever APIs).
- Go to Settings → Profile and set an API password. This is a
dedicated password used only by API clients; it is separate from your login
password. Put it in
FRESHRSS_API_PASSWORD.
Running locally
cp .env.example .env # then edit .env with your instance details
pip install -r requirements.txt
set -a && . ./.env && set +a
python server.py
# MCP endpoint: http://localhost:8080/mcp
# Health check: http://localhost:8080/healthz
Or with Docker:
docker build -t freshrss-mcp .
docker run --rm -p 8080:8080 --env-file .env freshrss-mcp
Connecting a client
Add it as a streamable-HTTP MCP server. With the Claude Code CLI:
claude mcp add --transport http freshrss https://your-host/mcp \
--header "Authorization: Bearer $MCP_TOKEN"
Any MCP client that supports HTTP transports works the same way: point it at
https://your-host/mcp and send Authorization: Bearer <MCP_TOKEN>.
Security notes
- No secrets live in this repo. Credentials are supplied only at runtime via
environment variables.
.envis gitignored; only.env.example(placeholders) is committed. - Always set
MCP_TOKENfor any networked deployment — without it the endpoint is open to anyone who can reach it. - The server stores no data on disk; every call proxies to FreshRSS. The ClientLogin token is held only in memory and refreshed automatically on 401.
- Prefer giving the MCP its own FreshRSS user/API password so access can be revoked independently of your main account.
How it works
server.py is a single FastMCP
streamable-HTTP app. A small GReader client handles ClientLogin auth (token
cached in memory, auto-refreshed on 401) and the handful of GReader endpoints
the tools need. A Starlette middleware enforces the bearer token on everything
except /healthz.
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.