GitHub: https://github.com/pangon/ai-sdlc-scaffold
I built an open-source repository template for AI-first software development, designed primarily around Claude Code, focused on the pre-coding phases (objectives elicitation, user stories, requirements definition, and design) that developers tend to cut short under tight time constraints. AI agents can help a lot in these phases too, not just in writing code.
Core principles:
Everything-in-repo: objectives, user stories, requirements, architecture, decisions, and task tracking all live alongside the source code.
Context-window efficiency: instructions and project knowledge are organized hierarchically so the agent can load only what is needed. Artifact collections are indexed via markdown tables.
Decision capture: decisions made during AI reasoning are captured and persisted as structured artifacts in the repo, so they remain reviewable, traceable, and consistently applied across sessions.
I have been using this approach for my personal projects for a while and decided to package it up in a way that might be useful to the community.
Licensed under Apache 2.0 — fork it, adapt it, build on it.
Feedback, criticism, and contributions are very welcome. I'd love to hear what works, what doesn't, and what you'd change.
Top comments (0)