Daemons by Charlie Labs
Keep PRs, issues, CI, and docs moving with AI agents
587 followers
Keep PRs, issues, CI, and docs moving with AI agents
587 followers
Charlie Labs gives engineering teams always-on AI daemons that keep work moving after coding agents create it. Define recurring roles in your repo, then let Daemons monitor PRs, issues, CI, docs, and Sentry errors over time. Instead of waiting for another human prompt, Daemons leave reviewable updates where your team already works: GitHub, Linear, Slack, and Sentry.
Free Options
Launch Team / Built With
Amazing! I have tinkered with self maintaining docs earlier and this looks like a much cleaner way to give more control on when the tasks are invoked. I am assuming teams could choose to make them a CI/CD gate or choose to do retrospective cleanup? Can they be customized beyond the .md to run custom scripts etc?
AISA AI Skills Test
the 'agents create work' framing is spot on. every team I've seen go heavy on coding agents ends up drowning in the review/triage overhead nobody planned for. curious how you handle the trust question — do teams typically start with daemons on low-stakes repos first, or jump straight in?
I like that Daemons by Charlie Labs focuses on a specific workflow instead of trying to be a broad all-in-one tool. The part I would want to understand better is the first-time setup: how quickly can a small team get from a blank workspace to a useful result?
The useful version of this is probably less about “agents do everything” and more about keeping the boring handoffs from stalling: stale PRs, missing context, failed CI, outdated docs.
How do you avoid the agent creating more notification noise than it removes?
The markdown-defined roles are the part I’d test first, especially for launch repos where docs drift and CI flakes pile up after a coding agent opens PRs. I like that the daemon can no-op when nothing changed instead of becoming noisy. If a daemon proposes a fix that touches a protected file or customer-facing doc, can teams require a human approval rule before it opens or updates the PR?
Strong thesis. Most teams are focused on code creation, not the maintenance burden that comes after.
Daemons by Charlie Labs
@oreofe_oluwatipin1 it aint much but it's honest work 🤠 🧹
Just ran the terminal prompt—under 60 secs is no joke. Agent replied on Teams and held context when I responded. Solid.