AI-DLC Workflows 2.4 · Blueprint v3

AI-Driven
Development Lifecycle

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.

32
Stages
11
Domain agents
9
Scopes
4
Harnesses
⓪ 3 stages
Initialization
bootstrap workspace & state
① 7 stages
Ideation
validate intent, define scope
② 8 stages
Inception
requirements, architecture, plan
③ 7 stages
Construction
design, implement, test
④ 7 stages
Operation
deploy, observe, optimise

How it works

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.

Process

The whole stage graph as a single spine, with inputs, outputs and gate rules.

Harness

The work, the mind, the machine, and the dials and verbs that control it.

Artifacts

All 57 artifacts with templates, worked examples and lineage.