Digest

A fast, lightweight RSS/Atom feed reader with both CLI and MCP (Model Context Protocol) server interfaces.

Features

Feed Management

• Add feeds with optional folder/category organization
• Remove feeds (cascades to delete all entries)
• Move feeds between folders for reorganization
• Auto-discover feed URLs from website URLs (built into feed add)
• OPML import/export for feed subscriptions

Entry Tracking

• Fetch feeds with HTTP caching (ETag, Last-Modified) for efficiency
• List entries with filtering by feed, category, read status, and date
• Smart date filters: today, yesterday, week, month
• Read articles with HTML-to-markdown conversion
• Mark as read/unread - individual entries or bulk by date

Storage Backends

• SQLite - fast, full-featured with FTS5 full-text search
• Markdown - human-readable file-based storage via mdstore
• Configurable via digest setup interactive wizard

MCP Server

Full MCP integration for AI agents to manage feeds:

MCP Resources

MCP Prompts

Workflow templates for common RSS management tasks:

Example prompt usage:

# Morning digest
Use the daily-digest prompt to catch up on today's news

# After vacation
Use the catch-up prompt with days=14 to process two weeks of entries

# Feed cleanup
Use the curate-feeds prompt to optimize my subscriptions

Installation

# From source
go install github.com/harper/digest/cmd/digest@latest

# Or clone and build
git clone https://github.com/harperreed/digest.git
cd digest
make build

CLI Usage

# First-time setup (choose storage backend and data directory)
digest setup

# Add a feed (auto-discovers feed URL from HTML pages)
digest feed add https://example.com/feed.xml
digest feed add https://example.com --folder "Tech"
digest feed add https://example.com --title "My Blog" --no-discover

# List feeds
digest feed list

# Move feed to category
digest feed move https://example.com/feed.xml "News"

# Remove a feed
digest feed remove https://example.com/feed.xml

# Manage folders
digest folder add "Tech"
digest folder list

# Fetch new entries from all feeds
digest fetch
digest fetch --force              # Ignore cache, force re-fetch
digest fetch https://example.com  # Fetch single feed

# List entries
digest list                    # Unread entries (default limit: 20)
digest list --all              # Include read entries
digest list --today            # Today's entries
digest list --yesterday        # Yesterday's entries
digest list --week             # This week's entries
digest list --category "Tech"  # Entries from Tech folder
digest list --feed <url>       # Entries from a specific feed

# Read an article (supports ID prefix matching)
digest read abc12345
digest read abc12345 --no-mark    # Read without marking as read

# Open article link in browser
digest open abc12345

# Mark as read
digest mark-read abc12345              # Single entry
digest mark-read --before yesterday    # Bulk mark

# Mark as unread
digest mark-unread abc12345

# Export data
digest export                      # OPML to stdout
digest export --format yaml        # Full YAML export
digest export --format markdown    # Markdown export

# Migrate between storage backends
digest migrate

MCP Server Usage

Configuration

Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "digest": {
      "command": "/path/to/digest",
      "args": ["mcp"]
    }
  }
}

Example Agent Workflows

# Get today's news
list_entries { "since": "today", "unread_only": true }

# Read an article
get_entry { "entry_id": "abc12345" }

# Organize feeds
move_feed { "url": "https://example.com/feed", "folder": "Tech Blogs" }

# Catch up on old articles
bulk_mark_read { "before": "week" }

Data Storage

• Config: ~/.config/digest/config.json
• Data directory: ~/.local/share/digest/ (respects XDG_DATA_HOME)
• Subscriptions: ~/.local/share/digest/feeds.opml (OPML)

Development

# Run tests
make test

# Build
make build

# Install locally
make install

# Run MCP server
./digest mcp

License

MIT