Self-Driving Agentic SDLC Factory

// CASE RECORD

The problem

Modern coding agents produce real work, but every handoff still wants a person: decide what starts next, keep the repository from overloading, notice a blocked task, run a review pass, route the fixes back, merge what’s finished. That orchestration is exactly the mechanical overhead that pulls a developer out of the creative, high-leverage part of the job — and it caps how much an agent fleet can actually ship.

What I built

This is my own system, and I run it in production. I made the conductor itself programmable and autonomous, with an agent as the worker behind every step, so a prioritized backlog becomes a living assembly line: issues → implementation → PR → self-review → fixes → squash merge. It picks up each unblocked issue, implements it, opens a pull request, reviews its own work, fixes review comments, merge conflicts, and failing CI, then squash-merges with closing references preserved — and moves on to the next. A manual label keeps any issue or pull request under direct human control. I stay on intent and architecture; the loop handles the handoffs, rebases, and CI fixes in the background, so only finished work accumulates.

Where it stands

Scaling. I measure it on autonomous merge rate — issues taken from open to merged without human intervention — plus human touch-points per shipped change and CI-fix turnaround, the time from a failing check to a pushed, passing fix. The pattern generalizes: any repository with a well-formed backlog can adopt the same self-driving loop, which makes this a reusable delivery-operations capability rather than a one-off script.