Top AI Product

Coding agents are great at building, but they leave a mess behind — outdated dependencies, dead code, redundant abstractions. Daemons, a new product from Charlie Labs, is built to clean that up continuously: the team’s framing is “use agents to build, and use daemons to maintain what you’ve built.”

## How Daemons work

A daemon is defined in a Markdown file (DAEMON.md) that lives in your repo, describing a recurring job. Daemons then run always-on, watching for triggers like a pull request merging into the default branch or a security advisory landing for one of your dependencies. From there they handle the ongoing loops: opening tested dependency-upgrade PRs, patching known vulnerabilities, removing dead code and unused exports, and cleaning up the redundant abstractions that pile up from heavy agent use.

## Why it is a distinct category

Charlie Labs pitches daemons as separate from one-shot agents: agents do novel creative work, daemons keep the recurring maintenance moving after everyone has moved on. A shared token budget covers daemon work, so each engineer’s allocation goes toward new features rather than background upkeep. For teams whose agents now generate a lot of code, an always-on maintenance layer that keeps the repo patched and tidy is the unglamorous counterpart to all that generation.