Comparisons · Updated 2026-11-03
Spec-driven vs outcome-driven development
Spec-driven development (Spec Kit, Superpowers, BMAD, Kiro and others) governs how an agent turns intent into code. Outcome-driven development governs whether that code was worth it. They are complements: use a spec tool to build, and an outcome loop to decide what to build next.
What spec-driven tools do well
They make the plan a persistent artifact, so an agent does not drift between sessions, and they put a human review between spec, plan and code. For long builds that is a real improvement over prompting.
Where they stop
At the merge. None of them asks what number the change should move, ships it behind a switch, or records whether it worked. That work lands in separate analytics and flag tools that never see the spec.
Where Golden Frijoles fits
It runs the whole practice in one loop your agent drives: shape the bet, build and review it, ship it behind a flag, measure it, record the verdict. If you already like a spec tool for the build step, keep it.
Questions
Do I need to stop using Spec Kit or Superpowers?
No. They cover the build; an outcome loop covers what happens after the merge.