Claude Code on a compiler for weeks at a time, with a human checking in
Reviewing every turn kills the value. Working from a spec and a to-do list, Claude Code runs for weeks on a compiler and a strategy game while they step in now and then.
Done withClaude CodeSpotted· by HN user vidarh· Oct 17, 2026
Result A spec and a to-do list keep it on track
“I have multi-week runs of Claude Code going to work on a compiler project. I have a week-long run of Claude Code where it is writing a real-time strategy game.”
Spotted: found in the wild. Details are as reported by the source above and have not been independently verified.
What they did
They set Claude Code to work against a written spec broken into a plan and TODO list; for the compiler the spec came from an external test suite, and for the strategy game Claude Code wrote its own spec first. They keep a multi-week run going on the compiler and a week-long run going on the game.
How it went
They only step in every few tens of minutes to a few hours to review code and nudge it back on track, with a closer review every day or so. They say the time savings come specifically from needing a human in the loop less often, and that reviewing every single turn would drastically reduce the value.
Try it yourself with Claude Code
Give your coding agent a written spec and a generated TODO list for a self-contained project, then let it run unattended for stretches of time rather than reviewing every step. Check in every so often to review its code and redirect it when it has drifted, with a deeper review roughly once a day, and see how long a run you can sustain before intervention stops paying off.
A starting prompt written by Agents in the Wild. Fill in the [brackets], and check what the agent does before it sends, pays for or deletes anything.
Markdown files should just open, so they asked Claude Code for a viewer. It opens instantly: faster, they say, than any other macOS app, and faster than Bitwarden's Chrome extension shows its fill menu.
Migrating a password reset flow, Claude Code and Opus produced code on par with a mid level developer. It needed close supervision throughout, and cost roughly 2 months of Pro subscription for about half a day's work.
A URL shortener side project ran on AWS. Claude Code wrote the GitHub Actions workflows for the move and fixed each failure as it came up. The writer reports zero downtime.
A 7-app monorepo needed an 8 GB memory setting and 25 seconds to typecheck. Claude Code made the changes for the move to oxlint, oxfmt and tsgo, and the team validated each step.
Turning plan files into Claude Code workflows took 5-6 hours each. A script that converts them deterministically, and flags what can move out of the LLM, cut that to 20-30 minutes at about 20% of the tokens.
An inherited app on Azure, an error, and a hunt that could easily take 1-2 days by hand. With CLI access to Azure and the repo, Claude Code ran the config and network checks and found the root cause in about 45 minutes.