An open-source methodology from AWS Labs that gives AI coding agents a disciplined delivery process — defined artifacts, assigned agents, and a verification gate at every phase boundary.
Instead of letting a model generate code in one shot, AI-DLC breaks delivery into five phases. Each stage declares its inputs, its outputs and the agent that leads it, and every phase boundary is guarded by a gate.
A CLI like Claude Code or Kiro is already a harness for the model. AI-DLC builds a second harness on top: a methodology engine that drives the model through a controlled, auditable process.
The whole stage graph as a single spine, with inputs, outputs and gate rules.
The work, the mind, the machine, and the dials and verbs that control it.
All 57 artifacts with templates, worked examples and lineage.