h4cker turns Hacker News into a focused AI-native reading workspace. Browse the original feed, then use a persistent Flue-powered Agent to summarize stories, analyze discussions, research linked sources, and personalize what deserves your attention. First Brief, HN Scout, long-term reading memory, feedback-driven recommendations, and scheduled Agent Digests form a complete Agent loop. The code is publicly available on GitHub under a noncommercial source-available license.

Hey Product Hunt! I built h4cker because I wanted to learn Flue—an agent framework built on Pi Agent—through a small but complete real-world product. Hacker News felt like the perfect test bed: the source material is public, discussions are rich, and the daily reading workflow is easy to understand. What started as a faster HN reader grew into a persistent Agent system with First Brief, HN Scout, deep research, long-term reading memory, explicit feedback, and scheduled Agent Digests. Under the hood it uses Flue workflows, typed tools, specialized skills, and bounded read-only subagents, while keeping product data and runtime state clearly separated. The source is available on GitHub under a noncommercial license. I’d love feedback on whether the Agent actually helps you find and understand higher-signal technical discussions.

the fact that the agent actually remembers what you care about across sessions is kind of brilliant, like it feels less like a chatbot bolted onto HN and more like a real reading companion that learns your taste

h4cker.app

An AI-native reading workspace for Hacker News

An AI-native reading workspace for Hacker News