AI-DLC Workflows v2.11 · Claude Code

AWS AI-DLC, Step by Step

The AI-Driven Development Lifecycle: AI proposes, people approve at each gate, and every decision is kept in a record next to the code. Learn the ideas, then run one real requirement through it on a Spring Boot repo.

Start here: set up an AI-DLC workspaceFour commands, every file they create, and what each one is for: the team's files, the 314-file harness in .claude/, the record in aidlc/, and the code repo. Windows and macOS; real output.Starter: github.com/varasrinivas/aidlc-java-workspace (branch main) Hands-on run: one requirement, 17 steps (Java)From a Slack message to a pull request with the express profile, checked against a real run: what you type, what Claude returns, what lands on disk. Windows and macOS commands; real code, tests and output.Starter repo: github.com/varasrinivas/aidlc-priorauth-api The same run in Python (FastAPI)The 17 steps on a FastAPI repo managed with uv: same requirement, gates and record. Real code, tests and output; Claude's side carried over from the Java run.Starter repo: github.com/varasrinivas/aidlc-priorauth-api-python Variant: start from a Rally featureThe next feature on the same API, with a Rally feature and a SharePoint spec as inputs: what changes in the workspace and at each gate.Real input files and real aidlc output; no AI run replayed Feature scope: a real run through IdeationThe next request on the same API run with /aidlc feature instead of express: the reviewer agent, sensors, summary confirmations, learnings, feasibility, and what each stop cost.A real AI-DLC 2.11 run, stopped at 1.4 Scope Definition Concepts, explainedPhases, the 33 stages, profiles, gates, spaces and intents, the audit record, with diagrams. One-page cheat sheetThe same concepts on a single page, to keep open while you work. Concept mapHow the pieces connect, in one diagram. Starter repopriorauth-api at v1.0.2, the code before the change. Tag v1.1.0 is the answer key.

Suggested order: the concepts page (or the cheat sheet if you are short on time), then set up a workspace, then the hands-on run and its Rally variant.