Skip to main content
RStack is developed through research-backed design decisions, not ad-hoc feature copying. The project compares prior art, captures claims discipline, records RFCs, and then implements validated roadmap slices.

​
Development loop

research → issue → RFC/ADR → implementation PR → validation → evidence → paper update

​
What this means in practice

  1. Research first — RStack starts with references such as NIST Artificial Intelligence Risk Management Framework (AI RMF), NIST Secure Software Development Framework (SSDF), ISO/IEC 42001:2023 AI management systems, OWASP Top 10 for Large Language Model Applications, Supply-chain Levels for Software Artifacts (SLSA), Dead Simple Signing Envelope (DSSE), Sigstore, Augment Code’s AI-SDLC reference architecture, and ai-sdlc-framework/ai-sdlc.
  2. Issue tracking — roadmap ideas are captured in GitHub issues with acceptance criteria and paper angles.
  3. RFC/ADR — major design decisions are recorded under rfcs/ before implementation.
  4. Implementation — code and docs changes are delivered through PRs with local validation and CI.
  5. Evidence discipline — productivity claims are measured or clearly marked as hypotheses.

​
Primary-source artifacts

​
Current roadmap chain

The first research-backed chain is:
  1. #77 Research bibliography and methodology — completed first to establish claims discipline.
  2. #76 RFC / ADR process — records decisions before major implementation.
  3. #70 Decision Queue / DoR — first product feature that should follow an RFC.

​
Paper-safe wording

Use this wording until measured evidence exists:
RStack creates the governed lifecycle, evidence structure, and observability needed to measure and improve AI-assisted software delivery.
Do not claim quantified productivity improvement until RStack has measured comparison runs.