Lurk
Pull full Reddit threads and comment trees into agent context with far fewer tokens than raw JSON or HTML scraping.
Overview
Lurk is a MCP server for the Idea phase that reads Reddit threads and full comment trees into LLM-friendly form with about 94% fewer tokens.
What is this MCP server?
- Reddit reader optimized for LLMs with full nested comment trees
- Claims roughly 94% fewer tokens versus naive full-thread ingestion
- Single-binary style delivery via npm package reddit-lurker at v1.0.2
- Stdio MCP transport for Claude Code, Cursor, and similar clients
- Purpose-built lurker workflow instead of generic web browser automation
- npm package reddit-lurker version 1.0.2
- Vendor description: ~94% fewer tokens for full comment trees
- Transport: stdio; repository: github.com/ProgenyAlpha/reddit-lurker
Community signal: 2 GitHub stars.
What problem does it solve?
Reddit research in agents either misses nested comments or burns the whole context budget on bloated thread dumps.
Who is it for?
Indie founders doing audience and problem research on Reddit inside Claude Code or Cursor before writing specs or landing pages.
Skip if: Teams needing official Reddit API compliance reviews, paid social analytics, or non-Reddit community sources.
What do I get? / Deliverables
You register Lurk and get compact, tree-complete Reddit reads your agent can summarize for validation and positioning work.
- LLM-optimized Reddit post and comment-tree payloads
- Lower token footprint versus unstructured thread dumps (vendor claims ~94% reduction)
- Repeatable agent workflow for subreddit and thread deep dives
Recommended MCP Servers
Journey fit
Early journey work depends on hearing how real users talk about problems, competitors, and niches on public forums. Reddit deep reads are classic idea-phase research for pain points, language, and competitive chatter before you commit to a build.
How it compares
Token-efficient Reddit MCP reader, not a broad web-scraper skill or SEO rank tracker.
Common Questions / FAQ
Who is Lurk for?
Solo builders and indie hackers who use AI agents for market and audience research on Reddit and want full comment context without maximal token waste.
When should I use Lurk?
Use it in idea and validate research when you need verbatim community discussion, nested replies, and pain-point mining from specific subreddits or posts.
How do I add Lurk to my agent?
Install the reddit-lurker npm package v1.0.2, add it as a stdio MCP server in your client configuration, and invoke its Reddit read tools from your research prompts.