April 18, 2026
You sleep, it ships? Commenters disagree
Show HN: Remoroo. trying to fix memory in long-running coding agents
Overnight code butler or cringe? Devs roast the tagline and ask if it’s just another clone
TLDR: Remoroo says it improves your code overnight by testing changes and keeping what works. Commenters roasted the dramatic marketing, claimed they can do the same with tools like Claude or DIY projects, and demanded comparisons to open-source options like ralph—pushing for real proof over flashy agent hype.
Remoroo promises a dream: you go to bed, it runs experiments on your code, and by morning it’s kept the wins and tossed the flops. But the comments? They woke up spicy. One early take dragged the website’s dramatic slogans as “so painful to read,” quoting lines like “Not a coding agent. An autonomous research engine” and the mic-drop “It didn’t guess. It proved.” Cue eye-rolls and riffs on the boastful “You slept through it.”
Beyond the copy roast, the crowd split into camps. DIY builders rolled in, with one dev sharing myagent, a project that reads research papers and tries to implement them, framing this “overnight brain” idea as a fun weekend hack rather than a product. Meanwhile, a power user said they don’t need a new service because work gives them access to a top-tier chatbot (Claude), and they already ask it to do this stuff—translation: why pay for what a smart assistant can do?
Then came the accountability check: another commenter asked how Remoroo differs from ralph, an open-source cousin. That sparked the bigger mood: is this real step forward or just shiny agent déjà vu? In short, the tool’s pitch is bold, the results sound nice, but the community wants receipts—and fewer dramatic taglines.
Key Points
- •Remoroo runs autonomous research on code locally and overnight.
- •Users provide a specification file (e.g., program.md) to guide experiments.
- •The tool edits code, runs tests, evaluates results, and keeps or reverts changes accordingly.
- •The goal is to deliver improved, validated results by morning with minimal supervision.
- •It is presented as addressing memory issues in long-running coding agents.