rss-mcp
Summary: A local-first RSS reader MCP server for feed management, clean article extraction, reading triage, and text/vector search.
Subscribe/unsubscribe/list/refresh feeds, fetch/get items, mark read/starred, remove items, import/export OPML, and audit feed health.
Extract clean article text with caching using Trafilatura/readability fallback, retrieve cached extracts, clear cache, and list extractors.
View inbox/starred items, run FTS5 full-text search and LanceDB semantic search, see stats/RAG stats, and trigger/check background indexing sweeps.
Get interactive help, render a Prefab reader stats card, and request graceful shutdown.
Run purely locally with SQLite, no account required; supports stdio/Claude Desktop and a separate webapp/frontend workflow.
Integrates with Ollama as an LLM provider for a built-in chat and exploration assistant, with dynamic provider detection for local LLMs.
Provides tools for subscribing to and managing RSS/Atom feeds, fetching and reading articles, full-text and semantic search, OPML import/export, feed health checks, and readability extraction.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@rss-mcp subscribe to https://news.ycombinator.com/rss and show me unread"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
rss-mcp - Superduper local-first RSS reader + readability extractor
Subscribe to feeds, get clean article text without ad sites, full-text + semantic search. Pure local SQLite, no account needed.
Features & Preview
Dashboard: High-level KPIs (feeds, total items, unread, starred).
3-Pane Reader: Feeds navigation, articles stream, and clean readability text extract pane.
Feeds Manager: OPML import/export, feed health audits, refresh triggers.
Inbox & Starred: Unified triage inbox and starred bookmarks.
RAG & Search: FTS5 SQLite full-text search combined with LanceDB semantic vector sweeps.
Chat & LLM: Built-in LLM exploration assistant with dynamic provider detection (Ollama, LM Studio, etc.).
Related MCP server: feedkit
Framework Stack
Backend: FastMCP 3.1+, Starlette / Uvicorn, SQLite3 WAL + FTS5, Trafilatura, LanceDB, Pyright (zero errors).
Frontend: React 19, TypeScript, Vite, Tailwind CSS, Zustand, Lucide React, Biome linter.
Install & Run
Option | Command |
A naked PC |
|
B dev backend |
|
C stdio (Claude Desktop) |
|
D frontend |
|
See INSTALL.md. Onboarding: N/A (pure local util, no wrappee/account) - rationale in docs/DEVELOPMENT.md.
Claude Desktop Configuration
Add to your claude_desktop_config.json:
{
"mcpServers": {
"rss-mcp": {
"command": "uv",
"args": [
"--directory",
"d:/Dev/repos/rss-mcp",
"run",
"python",
"-m",
"rss_mcp"
]
}
}
}Environment Variables
Variable | Default | Description |
|
| Backend server port (strict fleet standard) |
|
| Webapp frontend port |
|
| Host address to bind to |
|
| SQLite database file path |
|
| LanceDB vector database directory |
Tools (all real, no stubs)
feeds_ops: subscribe, unsubscribe, list_feeds, refresh_feed, refresh_all, fetch_items, get_item, mark_read, mark_star, remove_item, opml_import, opml_export, feed_healthreadability_ops: extract, get_cached, clear_cache, list_extractorsreader_ops: inbox, starred, search_text, search_semantic, stats, trigger_sweep, sweep_statusshow_reader_prefab_card: Prefab card view for visual readersrss_help: Interactive help and quickstartrss_shutdown: Clean server termination
Ports
Backend 11974, frontend 11975. Forbidden ports are never used (3000, 5000, 5173, 8000, 8080).
Install (.mcpb for Claude Desktop)
Claude Desktop: install
rss-mcp.mcpbfrom the latest GitHub release (releases/latest/download/rss-mcp.mcpb), then runinstall.ps1.Other clients (Cursor/Codex/CLI):
uvx rss-mcp(stdio transport, same tools as the bundle).
vs glance-mcp
glance-mcp is a lightweight holdall (single-shot fetch, weather, probes). rss-mcp is the stateful full reader: subscriptions, polling with ETag, readability cache, FTS5, stars, OPML, background sweeps.
Development
Justfile shortcuts:
just lint,just typecheck,just test,just ci,just gates-greenBiome check:
bun run biome:ci(inwebapp/)Pre-commit hook:
powershell.exe -NoProfile -File scripts/pre-commit-biome.ps1Pack MCPB bundle:
powershell.exe -NoProfile -File scripts/mcpb-pack.ps1
Available Tools
6 toolsfeeds_opsB
feeds_ops - Full feed subscription lifecycle.
RATIONALE: groups subscribe/unsubscribe/list/refresh/fetch/get/mark/star/remove/opml/health into one discoverable tool. Stateful reader, unlike glance-mcp single-shot fetch.
Return Format
Dictionary with success (bool), message (str), and operation-specific payload (e.g. feeds, items, feed, opml).
Examples
{"operation": "list_feeds"} -> lists all subscribed feeds.
{"operation": "subscribe", "url": "https://news.ycombinator.com/rss"} -> subscribes to RSS feed.
{"operation": "refresh_all"} -> polls all feeds for new articles.
{"operation": "fetch_items", "limit": 20, "unread_only": true} -> retrieves unread articles.
Inputs: operation (required), url, limit, title, +6 more. Returns: object
readability_opsA
readability_ops - Clean article text without ad sites.
RATIONALE: trafilatura primary plus readability-lxml fallback, cached in SQLite. No paywall bypass: if the page needs login, extraction reports failure honestly.
Return Format
Dictionary with success (bool), message (str), and article text fields (title, author, text, html, extractor, cached).
Examples
{"operation": "extract", "item_id": 3} -> downloads article and extracts cleaned text.
{"operation": "get_cached", "item_id": 3} -> retrieves previously cached clean extraction.
{"operation": "list_extractors"} -> lists available readability engines.
Inputs: operation (required), force, item_id. Returns: object
reader_opsA
reader_ops - Daily reading surface: inbox, stars, search, stats, RAG sweeps.
RATIONALE: single entry point for what a human checks every morning.
Return Format
{success, message, items/stats/job/job_id/status/chunks}
Examples
{"operation": "inbox"} unread items
{"operation": "search_text", "query": "vienna"} FTS search
{"operation": "search_semantic", "query": "rocket engine"} vector search
{"operation": "trigger_sweep", "mode": "incremental", "hours": 4} background index job
Inputs: operation (required), ctx, mode, hours, +4 more. Returns: object
rss_helpB
rss_help - Capability summary and op list.
Return Format
Dictionary with success (bool), message (str), registered tools, ports, and comparison vs glance-mcp.
Examples
{} -> summary of reader capabilities and registered tools.
Inputs: no named parameters. Returns: object
rss_shutdownA
rss_shutdown - Graceful shutdown hook for the daemon.
Return Format
Dictionary with success (bool) and acknowledgement message (str).
Examples
{} -> acknowledges shutdown request.
Inputs: no named parameters. Returns: object
show_reader_prefab_cardB
show_reader_prefab_card - Prefab stats card for list/status/stats coverage.
Return Format
ToolResult with PrefabApp structured content plus plain-text fallback.
Examples
{} -> renders reader stats card.
Inputs: no named parameters.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 6 tools
feeds_ops and reader_ops both handle item retrieval and marking (e.g., fetch_items vs. inbox, mark/star in feeds_ops vs. stars in reader_ops), creating overlap that could cause misselection. show_reader_prefab_card also overlaps with reader_ops stats. Descriptions clarify intended separation, but boundaries are not crisp.
Most tools use a consistent noun_ops pattern (feeds_ops, readability_ops, reader_ops), but rss_help and rss_shutdown use a different prefix, and show_reader_prefab_card uses a verb_noun format. The mix is readable but not predictable.
Six tools is well-scoped for an RSS reader daemon. Each tool has a clear role: feed lifecycle, extraction, reading surface, help, shutdown, and a presentation card. None feels redundant or excessive.
The surface covers the full RSS reader lifecycle: subscription management (subscribe/unsubscribe/list/refresh/remove/OPML/health), article extraction and caching, reading inbox/stars/search/stats, background RAG sweeps, help, and graceful shutdown. No obvious operational gaps remain.
Maintenance
Related MCP Connectors
Read, search, star and organize your RSS feeds, newsletters and saved articles.
RSS & Atom feed manager with full-text search, AI curation, and remote MCP endpoint.
Read, search and track your RSS feeds: semantic search, story clustering, watches, OPML import.
Managed cloud RSS and Atom feed reader with a remote Model Context Protocol (MCP) endpoint. Zero server setup required — enables AI agents like Claude to instantly subscribe to feeds, search full-text articles, extract readable content, and access curated AI editions ("Your Newspaper") over secure streamable HTTP.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables intelligent RSS feed management with AI-powered semantic search, advanced filtering, and a comprehensive reading workflow. Supports OPML parsing, article organization with status tracking, and token-efficient browsing of large feed collections.5MIT
- AlicenseNot gradedqualityDmaintenanceEnables users to search a curated RSS/Atom catalog, subscribe to and collect feeds, and query collected articles through natural-language tools.2MIT
- AlicenseAqualityCmaintenanceEnables querying a local SQLite database of curated RSS/Atom/RDF feeds through tools for latest headlines with filters, keyword search, cross-outlet coverage comparison, and source/freshness listing. Items are kept in their original language so an LLM can translate, summarize, and compare framing at query time.4Mozilla Public 2.0
- AlicenseNot gradedqualityBmaintenanceEnables searching, reading, and privately archiving news from supported publisher subscriptions, with per-user authentication via human sign-in links and separate tokens, sessions, and archives.321 PyPI1MIT