AI-DLC Workflows
github.com/awslabs/aidlc-workflows- Category
- AI Agents
- Rank
- No. 833Tools index
Previous survey · No. 823 ·
- Type
- APP
- Builder
- awslabs
- GitHub
- 4.6k stars
- Latest release
- v2.9.0
- Date
About
AWS's AI-Driven Life Cycle — adaptive workflow steering rules for AI coding agents.
What it does
AI-DLC gives a coding assistant a staged process for understanding, designing, building, and checking software. It records questions and decisions in Markdown, pauses for approval at stage boundaries, adjusts depth to project complexity, and writes application code outside its documentation area.
Why it's ranked here
Its strongest case is disciplined continuity across long agent sessions. Durable question files, approval gates, execution plans, and state tracking make decisions inspectable and resumable. That structure demands more preparation and operator attention than an informal coding chat.
What's good
It supports both new and existing codebases, with reverse engineering for the latter. Teams can supply vision, scope, technical constraints, prohibited technologies, security requirements, and example patterns. The workflow turns those inputs into requirements, designs, code plans, tests, and an audit trail.
Tradeoffs
The process produces substantial Markdown and expects humans to review each stage before continuing. Users must manage context carefully, answer questions in generated documents, and explicitly prevent exploratory questions from becoming edits. Model output and generated costs still require review.
How to use it well
Use it for projects where requirements, architecture choices, approvals, and traceability matter more than rapid improvisation. Prepare a concise vision and technical environment first, then reset context at natural gates. It complements a coding agent; it does not provide the model, editor, deployment platform, or independent human assurance.
Technical notes+
README.md documents release-zip installation and agent-specific rule placement. docs/WORKING-WITH-AIDLC.md defines file-based clarification, approval gates, context resets, depth controls, and separate treatment for greenfield and brownfield work. docs/GENERATED_DOCS_REFERENCE.md catalogs conditional Markdown artifacts across inception, construction, build-and-test, and operations. docs/writing-inputs/inputs-quickstart.md specifies the minimum vision and technical-environment inputs. docs/DEVELOPERS_GUIDE.md and docs/ADMINISTRATIVE_GUIDE.md describe CodeBuild-based evaluation, release automation, pull-request gates, and six security scanners used by the repository.
Observed
- Primary format
- Markdown rule files and generated Markdown documentation
- Packaging
- Distributed through GitHub Releases as a zip containing core and detailed rule directories
- Install surface
- Project-local rule or steering files copied into the coding assistant's expected configuration area
- Supported coding assistants
- Kiro, Amazon Q Developer, Cursor, Cline, Claude Code, GitHub Copilot, and OpenAI Codex
- Operating systems
- Setup instructions cover macOS, Linux, and Windows
- Project modes
- Supports greenfield and brownfield projects, including reverse engineering for existing codebases
- Workflow interfaces
- Coding-assistant rules and Markdown approval artifacts; no standalone user CLI or library interface is described
Read from README.md, docs/DEVELOPERS_GUIDE.md, docs/WORKING-WITH-AIDLC.md, docs/ADMINISTRATIVE_GUIDE.md, docs/GENERATED_DOCS_REFERENCE.md, docs/writing-inputs/inputs-quickstart.md, docs/writing-inputs/vision-document-guide.md, docs/writing-inputs/technical-environment-guide.md.
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.