@harrypotter482
Founder & Software Developer, ModuleFlow
45m ago · Vibe-coding · Quito

Which assumptions should block an AI coding agent?

An unanswered question doesn’t disappear when implementation starts. It can become an implicit decision.
Consider a hypothetical request: “Let users export their data.”

An agent could produce a working implementation while silently choosing who can export, which records are included, and what happens when access is denied. Even the tests might reinforce those assumptions.

I’m developing Rigorkit to make these gaps explicit before implementation: what’s known, what remains unresolved, and which decisions need clarification.

The distinction I’m interested in is:

  • Clarify first when the answer changes permissions, expected behavior, scope, or acceptance criteria.
  • Explore with a prototype when the uncertainty is better resolved by trying something small and reversible.

That’s the approach I’m aiming for, rather than requiring every detail to be settled upfront. How do you decide which unknowns should block implementation and which an agent can safely explore? github.com/harrypotter48/rigorkit

AgenticCodingSpecDrivenDevelopmentAIWorkflowRigorkit

Comments

No comments yet
Start the conversation.
Log in to join the thread.