SDD Catalog
Execution Workflow AWS Labs MIT-0

AWS AI-DLC Workflows

AI-Driven Development Life Cycle — a prompt/rules package that steers an AI coding agent through Inception, Construction, and Operations phases, emitting reviewable markdown artifacts at each human-gated stage. Not a CLI or MCP server: activated in-agent with the phrase "Using AI-DLC, ...".

Repository Documentation

Detection

A project uses AI-DLC when any of these exist:

  • aidlc-docs/aidlc-state.md
  • .aidlc-rule-details/
  • .kiro/aws-aidlc-rule-details/
  • .amazonq/aws-aidlc-rule-details/
  • **/aws-aidlc-rules/core-workflow.md

Status Model

Method: task-checkboxes

Parse the "## Stage Progress" checklist in aidlc-state.md (respecting SKIPPED as done/not-applicable), then refine per-unit progress from the checkbox ratio in each construction/plans/<unit>-code-generation-plan.md. Formats are LLM-generated markdown — match checkboxes plus the keywords COMPLETED/SKIPPED/IN PROGRESS rather than exact headings.

Lifecycle

  1. 1
    Inception

    Planning — WHAT & WHY. Workspace detection, (brownfield) reverse engineering, requirements analysis, user stories, workflow planning, application design, and units-of-work generation.

  2. 2
    Construction

    HOW — per unit of work: functional design, NFR requirements and design, infrastructure design, code generation, then build and test.

  3. 3
    Operations

    Reserved for deployment and monitoring (placeholder in current release).

On-Disk Artifacts

aidlc-docs/aidlc-state.md config Primary state tracker — project info, extension config, and the "## Stage Progress" checklist. Definitive marker that a run has occurred.
aidlc-docs/audit.md config Append-only, timestamped log of every interaction.
aidlc-docs/inception/requirements/requirements.md requirement Analyzed requirements — the hinge every downstream stage builds on. Maps to PRD.
aidlc-docs/inception/user-stories/stories.md requirement User stories and personas (conditional stage).
aidlc-docs/inception/plans/execution-plan.md plan Which downstream stages run, their depth, and a workflow visualization. Maps to PLAN.
aidlc-docs/inception/application-design/ design Components, methods, services, and dependency maps (conditional).
aidlc-docs/construction/<unit>/functional-design/ design Per-unit business logic, domain entities, and rules.
aidlc-docs/construction/plans/<unit>-code-generation-plan.md tasks Per-unit code-generation checklist (checkbox steps).
aidlc-docs/construction/<unit>/code/ output Markdown summaries of generated code (the code itself lands at the workspace root).
aidlc-docs/construction/build-and-test/ tests Build and test instruction files plus a summary.
./ output Application code, tests, and IaC written to the workspace root.

Fed by Definition Workflows

AWS Working Backwards → AI-DLC Definition side runs Amazon Working Backwards (aws-one-way-door) to produce the six-pager, PRD, and TRD; the execution side hands those to AWS AI-DLC, whose Requirements Analysis absorbs them as the Vision and drives Construction.