Searching Elfeed RSS Database

This skill provides access to the user's elfeed RSS feed database through a command-line interface. Elfeed is their primary information aggregation system, containing curated feeds from Reddit, blogs, YouTube channels, GitHub releases, and newsletters.

When to Invoke This Skill

Use this skill when:

• Topic research: User asks about topics likely covered in their feeds (programming, AI, security, NixOS, Emacs, etc.)
• Article recall: User references "something I read" or "an article about..."
• Pattern analysis: Understanding what they've been reading recently
• Context gathering: Building background on topics from their curated sources
• Unread triage: Helping prioritize or summarize unread items

Do NOT invoke for:

• General web searches (use WebSearch instead)
• Information not likely in RSS feeds
• Real-time news (elfeed may not have fetched yet)

Quick Start

The primary tool is elfeed-cli located at:

~/.claude/skills/elfeed/scripts/elfeed-cli

Or use the absolute path from this repo:

/home/markus/repos/nixos-config/nixos-shared/claude/skills/elfeed/scripts/elfeed-cli

Most common usage:

# Search recent unread items about a topic
elfeed-cli query "topic +unread @1-week-ago" simple 20

# Get database statistics
elfeed-cli stats

# List all feeds
elfeed-cli feeds

# List all tags
elfeed-cli tags

Filter Syntax Quick Reference

Combine these operators to build queries:

| Operator | Meaning | Example | |----------|---------|---------| | +tag | Must have tag | +unread, +programming, +reddit | | -tag | Must not have tag | -unread, -youtube | | @time | Newer than time | @1-week-ago, @3-days-ago, @1-month-ago | | #N | Limit to N entries | #10, #50 | | text | Title/link matches (regex) | nixos, claude.*ai | | !regex | Title/link doesn't match | !advertisement | | =regex | Feed URL/title matches | =simonwillison | | ~regex | Feed URL/title doesn't match | ~reddit |

Combine operators with spaces:

"+programming +unread @1-week-ago #20"
"+reddit -unread ai.*agents"
"+github =anthropic"

Common Workflows

1. Research a Topic

# Find recent articles about a topic
elfeed-cli query "topic-keyword +unread @2-weeks-ago" simple 30

2. Check Unread by Category

# Recent unread programming articles
elfeed-cli query "+programming +unread @1-week-ago" simple 20

# YouTube videos unwatched
elfeed-cli query "+youtube +unread" simple 10

3. Analyze Reading Patterns

# Get overall statistics
elfeed-cli stats

# See all feed sources
elfeed-cli feeds

# Check what tags are used
elfeed-cli tags

4. Find Specific Article

# Search by keyword across all entries
elfeed-cli query "specific-term" simple 50

# From specific source
elfeed-cli query "=simonwillison llm" simple 20

Output Formats

The query command supports three formats:

1. simple (default, most readable):

   2025-10-19 14:30 | Blog Name | Article Title | https://example.com/article
   

2. json (for parsing):

   [{"title":"...","link":"...","date":...,"tags":[...],"feed_title":"...","feed_url":"..."}]
   

3. sexp (Emacs s-expression):

   ((:title "..." :link "..." :date ... :tags (...) :feed-title "..." :feed-url "..."))
   

Important Notes

• Requires running Emacs: The CLI uses emacsclient to query the live database
• Real-time data: Queries reflect current database state (may include recently fetched items)
• Tag system: Feeds are tagged by category (programming, reddit, youtube, github, etc.)
• No side effects: Queries are read-only and don't modify read/unread status

Advanced Usage

For detailed filter syntax, regex patterns, and complex query construction, see reference.md.

Integration with Existing Command

The /mh:elfeed-search slash command is a convenience wrapper that calls this same tooling. You can invoke the skill autonomously or users can explicitly call the command.