April 21, 2026
Codebase Roomba or rebranded cron?
Show HN: Daemons – we pivoted from building agents to cleaning up after them
HN’s split: robot janitors for code or just fancy hooks
TLDR: A team unveiled “Daemons,” always-on AI helpers defined in simple Markdown files to keep code and issues tidy without prompts. The thread split between fans of the always-watching model and skeptics who say it’s just hooks or callable skills, with side-eye about competitors and how multiple daemons avoid collisions.
A startup showed off “Daemons,” always-on AI “robot janitors” that live in simple .md files and quietly tidy your code life: they watch pull requests, label issues, fix flaky checks, and even schedule their own rounds. The pitch: agents need a prompt, daemons just notice mess and clean it up. Think job description in Markdown, not another chat window.
Cue the HN peanut gallery. Skeptics came out swinging: one voice boiled it down to, why isn’t this just a callable skill? Others poked the hype balloon by asking how this differs from old-school hooks—those event-triggered scripts developers have used forever. The savviest analogy landed hard: it’s cron vs a running service, with daemons keeping context over time instead of firing once and forgetting. Meanwhile, competitive radar lit up with a pointed, “How does this compare to OpenProse? Additive or head-to-head?”
The nerdiest drama surfaced around drift detection and chaos control: what happens when two daemons want to touch the same files—do we get rules for who goes first, or do they each work in their own branches? Through it all, commenters joked about “hiring a tiny Roomba for your repo,” while others warned it could just be a glossy rebrand unless it truly manages long-running context and enforces guardrails. Verdict: intrigue meets side-eye, with everyone waiting to see if these daemons are guardians or just glamorized scripts.
Key Points
- •Daemons are self-initiated AI background processes defined in Markdown files to maintain code and workflows.
- •A Daemon’s frontmatter declares its name, purpose, watch conditions, routines, deny rules, and schedule; the body sets policy and limits.
- •Example ‘pr-helper’ Daemon improves PR descriptions and reviewer context while being denied merge/push permissions.
- •Example ‘issue-labeler’ Daemon adds missing labels to Linear issues under strict policies and processing limits.
- •Daemons target operational debt by monitoring tools like GitHub, Linear, Sentry, and Slack to detect drift and act without prompts.