Agent Rigor

Agent Rigor is an open-source skill repository that enforces engineering discipline in AI coding assistants by making them follow a strict six-phase development cycle. Each phase mandates empirical verification and testing, so the assistant has to prove its code works rather than declaring a task done after generating it. The goal is to fix a common failure mode — AI tools that produce plausible but untested or broken code — by structuring the workflow around checks at every step. Distributed on GitHub, Agent Rigor is a set of skills you add to a coding assistant rather than a standalone app, so developers can adopt the loop inside the tools they already use. It's best for engineers who want more reliable, well-tested output from AI pair programmers and are willing to trade some speed for verified results.

Developers use Agent Rigor to enforce a six-phase, test-driven workflow on AI coding assistants so they produce verified, reliable code.