Fixing a Live iOS App With an Auditable Software Factory

I built an auditable autonomous software factory to maintain a production iOS app, epac. Transparency is first-class: the app is open source, and the entire SDLC is public, from the provenance of feature ideas to the evidence that an implementation works, to the released version containing the change.

I got carried away with my new Claude Code Max plan. I handed Claude the epac repo, a note full of feature ideas, and instructions to create Linear items for each and implement them. I woke up to a feature-rich app that was submitted for review and a provisioned backend infrastructure that cost real money. Output velocity was high, but the work was not auditable - that’s where the software factory comes in.

In the demo, I will walk through the public artifacts produced by the autonomous SDLC as it shipped a bug fix (EPAC-1940): LLM session → Root Cause Analysis → Linear ticket → autonomous implementation → evidence bundle → TestFlight → App Store release. Then I will show any active work items in the factory, with provenance and evidence attached to each. I’ll open the autonomously reviewed PRs to show code diffs and review comments. OpenClaw acts as the Factory Foreman surfacing slowdowns (runaway turn count, stalled execution, human-required code review) and answering operator questions about the state of the factory. I’ll also highlight useful telemetry regarding token use, efficiency, and cost.