Show HN: Readhn – AI-Native Hacker News MCP Server (Discover, Trust, Understand)
TL;DR
Readhn is an open-source MCP server for Hacker News with three pillars: Discovery, Trust, and transparent ranking.
Key Points
- It ships 6 tools: discover_stories, search, find_experts, expert_brief, story_brief, and thread_analysis.
- An EigenTrust-style model propagates credibility scores outward from manually seeded expert accounts.
- After pip install, a single readhn setup command auto-configures the server in Claude Code, Codex, Cursor, and others.
- Every ranking result exposes its signals explicitly rather than returning an opaque score.
Nauti's Take
This is an honestly built project – no inflated pitch, just a real frustration turned into a working tool. The EigenTrust idea is not new, but applying it to HN specifically makes sense: genuine influence there is nearly invisible to outsiders.
The auto-configuration via readhn setup is the most practical detail – MCP server setup remains one of the biggest friction points for non-developers. What is missing is transparency around seed expert selection; whoever controls that initial list ultimately controls the trust model.
Context
Hacker News is one of the densest sources of early technical signals, yet it lacks any API for semantic search or expert filtering. Readhn closes that gap directly inside AI agent workflows instead of the browser. The trust model is the interesting part: anyone familiar with HN knows raw karma is a poor quality signal – EigenTrust propagation from curated seed accounts is a more realistic approximation of actual credibility.
Sources
