Ten minutes. That's how long developer Aaron Brethorst spent generating a user story that included 30 acceptance criteria, a full STRIDE threat analysis, a data classification matrix, and two Mermaid diagrams. The manual equivalent? Half a day, minimum - and he admits the diagrams simply wouldn't have existed.
Brethorst published a detailed walkthrough of a workflow he built using Claude Code's /skill-creator tool to automate user story generation. The setup runs parallel sub-agents acting as a Security Engineer and a Product Manager, each with role-specific review criteria. Conditional logic skips unnecessary reviews based on issue type. The system asks three targeted multiple-choice questions for clarification, drafts the story, then poses five open questions before filing the GitHub issue directly.
The Real Shift: Quality Work You'd Otherwise Skip
The interesting part isn't the speed. It's what happens when previously expensive work becomes free.
Brethorst's system caught an Android 11+ two-step permission flow that would have been easy to miss in a manual review. The STRIDE analysis (a structured way to identify security threats like Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, and Elevation of Privilege) surfaced security concerns that wouldn't normally enter a product discussion until much later.
As Brethorst put it: "Skills change the economics of quality. The cost of producing these diagrams dropped to zero, so now they exist."
That's the pattern worth paying attention to. Teams don't skip diagrams, threat models, and thorough acceptance criteria because they're unimportant. They skip them because the cost-benefit math doesn't work when you're shipping on a deadline. When AI collapses that cost to near-zero, the math flips.
What the Workflow Actually Looks Like
- Scaffolded with Claude Code's
/skill-creator, then tightened with multiple/simplifypasses - Two parallel sub-agents review from security and product perspectives
- Conditional logic routes issues to relevant reviewers only
- Auto-generates Mermaid flowcharts and sequence diagrams
- Collects human input through structured multiple-choice prompts
- Merges all findings, iterates once, then files directly to GitHub
This isn't a replacement for product thinking. The system still needs a human to answer clarifying questions and sanity-check the output. But it's a concrete example of how AI coding assistants are changing where teams spend their time: less on the mechanical work of writing acceptance criteria and drawing diagrams, more on deciding what to build and why.