A bug-fix automation must never exit quietly
The bug-fix automation in the software-factory.dev factory has two ways to end a run: it ships a fix or escalates to a human. Honest failure is a feature you have to design.
Written by
-
Benedikt Stemmildt Founder & Co-CEO
Published: 26 June 2026 Updated: 2 min read
There is one line in the Bug Fixer automation that I keep pointing people at. It sits in the prompt’s “do not” section, and it reads: never silently exit when something is broken.
The Bug Fixer runs in a software factory built in public at software-factory.dev. Its job sounds simple: pick the next open bug, fix it, open a pull request.
One run, step by step
It wakes every two hours, and there is a manual button for the moments someone wants to push it along. On each run it:
- checks that the environment is healthy
- counts the open pull requests and stops if three or more are already in flight
- picks the next bug in priority order and checks whether it already has a pull request open
- reads through closed issues to see if this one has shown up before, because a recurring bug usually means the last fix patched a symptom rather than the root cause
- reproduces the bug, writes the fix on a branch, runs the full test suite and opens the pull request
Two exits, no third
Now the detail worth copying. The Bug Fixer has exactly two ways to end a run. It ships a fix, or it cannot find the root cause and escalates the issue with a needs-human label and a comment explaining what it tried. There is no third outcome, no run where the work just quietly disappears.
That sounds small. I think it is the whole thing.
Loud failure is designed in
An automation you cannot watch all day is only trustworthy if its failures are loud. The dangerous version hits something it does not understand, returns nothing and reports success, and you find out weeks later that a whole class of bugs has been silently skipped.
Honest failure does not happen by default. You have to write it in, as plainly as this automation does, and then check that every exit path actually hits it. The full definition is in the Bug Fixer configuration on GitHub (opens in a new tab).
When your automations fail, does the failure show up somewhere a person will see it, or does it just end?
Further reading
Written by
-
Benedikt Stemmildt · Founder & Co-CEO
Factory skills, part 6 of 18
- The Feature Builder checks for open bugs before it builds The software-factory.dev build automation runs a pre-flight check first: a priority-1 or priority-2 bug nobody has picked up stops all new feature work.
- Daily metrics that cannot drift: subtract yesterday's total The Daily Metrics automation in the software-factory.dev factory never measures a daily change with a date-filtered query. It subtracts yesterday's cumulative total from today's, a method that cannot drift.
Scaling needs named owners
Coding agents scale across teams when named people own the platform, the team practice, the guardrails, and the measurement. Rollout & scale is where that ownership is built and handed to internal owners, in your stack.
Training on this topic: Agentic Software Factory: Advanced Team TrainingWant to check one team first? The self-check: twelve yes or no questions, three minutes, instant result, no account. Take the self-check