Pster's AI Workflow cuts repeat mistakes

This is less a prompt library than an operating model for AI-assisted engineering. The strongest idea here is that consistency comes from process enforcement, not from hoping a better model will remember your codebase.

• –The docs-first rule turns project documentation into working memory, which is a real fix for context loss across sessions
• –Splitting planning from execution makes it easier to keep architecture decisions stable before code starts changing
• –Optional quality gates let small tasks stay fast while still giving high-risk work more rigor
• –The workflow is open source, language-agnostic, and editor-flexible, so it can scale beyond one stack or one tool
• –The multi-root repo/workspace structure is a practical context win, especially when docs and code live in separate places