ADLC Framework
ADLC cuts delivery cost and risk while raising time-to-value. The Agent Development Lifecycle framework makes AI agent teams auditable and safe for enterprise. Coordinate with product owners and architects. Deploy with confidence.
For engineers
Enterprise AI Agent Governance, Testing & Deployment Orchestration
Version: 3.8.0
What is ADLC?
The Agent Development Lifecycle (ADLC) framework provides enterprise-grade governance for AI agent teams. It implements 7 constitutional principles with 92 enforcement hooks and 75 anti-pattern preventions to ensure safe, reliable, and auditable AI operations.
ADLC adds governance to your existing Claude Code workflow through agents, hooks, and skills. It works by adding a .claude/ directory to your project — there is no separate server or service to run.
Quick Start
# Install via npm
npx adlc@latest
# Or add as submodule to your project
git submodule add https://github.com/1xOps/adlc-framework.git .adlc
ln -s .adlc/.claude .claude
Key Features
| Feature | Description |
|---|---|
| Constitutional Agents | Product Owner, Cloud Architect, Security, and specialists |
| Marketplace Agents | Curated enterprise specialists across domains |
| Slash Commands | CDK, Terraform, K3D, FinOps, SpecKit workflows |
| Core Skills | Reusable capability packages with bundled references |
| Enforcement Hooks | Deterministic governance gates (tool-call level) |
| 3-Tier Testing | Snapshot → LocalStack → AWS Sandbox |
| Component Marketplace | Browse and install components |
Start with the ADLC Workflow page — it explains the mandatory coordination pattern that every agent and command depends on.
What Matters to You
For CEOs: Faster delivery, lower risk, auditable AI decisions.
For CTOs: Governance that scales, not bottlenecks. Standard patterns for AI coordination.
For Security/Compliance: Every agent action logged. Seven constitutional principles enforced. No rogue agents.
For engineers: Seven Principles
- Acceptable Agency - HITL approval for git, deployments, cost >$100/mo
- Interoperability - MCP standard, OAuth, least-privilege
- Evaluation-First - 100% code coverage, ≥95% agent behavior
- Hybrid Deployment - LocalStack (Tier 2) + AWS Sandbox (Tier 3)
- Observability - MELT telemetry, evidence logging
- Governance - 22 enforcement hooks before deployment
- Agent Engineering - Orchestrator pattern (50-100 LOC agents)
Next Steps
- Your Role, Your Path - Find the commands for your role
- ADLC Workflow - How coordination works
- Installation Guide - Get started in 5 minutes
- Browse Components - All components across agents, commands, skills and hooks
All ADLC components — agents, commands, skills, hooks, and MCP configurations — are browsable and installable from the Component Marketplace. You do not need to clone the full framework to use individual components.