Core Concepts
Understand how DaoFlow thinks about infrastructure, deployment, and agent safety.
Philosophy
DaoFlow is built on two core beliefs:
- AI agents are the future of infrastructure management — they should observe, plan, and execute deployments with proper safety boundaries
- Self-hosted infrastructure should be simpler than cloud — not more complex
Read the full Vision & Principles to understand why.
How It Works
Your Code → DaoFlow → Your Servers
↑ ↑
AI Agent SSH + Docker
DaoFlow sits between your code (or your AI agent) and your servers. It orchestrates Docker Compose deployments over SSH, tracks every change, and ensures agents operate within scoped permissions.
Key Concepts
| Concept | What It Is | Why It Matters |
|---|---|---|
| Architecture | Control plane + execution plane split | Separates orchestration from execution; Docker doesn't run in the web process |
| Projects & Environments | Organizational hierarchy | Group services by project, deploy to different environments (prod, staging, dev) |
| Servers | SSH-connected Docker hosts | DaoFlow manages your servers — any Linux box with Docker installed |
| Deployments | Immutable deployment records | Every deploy captures input, config, actor, timestamps, and outcome |
| Services | Runtime units (containers) | The thing that actually runs — from Compose, Dockerfile, or image |
| Vision & Principles | Why DaoFlow exists | Security-first, data ownership, transparency, deterministic by design |
The Safety Stack
What makes DaoFlow different from every other deployment tool:
┌─────────────────────┐
│ Approval Gates │ ← Human-in-the-loop for dangerous ops
├─────────────────────┤
│ Audit Trail │ ← Every mutation logged
├─────────────────────┤
│ API Lane Separation│ ← Read / Plan / Command
├─────────────────────┤
│ Token Scoping │ ← Per-token permission limits
├─────────────────────┤
│ Role Capabilities │ ← Role defines max permissions
├─────────────────────┤
│ Agent Principal │ ← Dedicated identity (read-only default)
└─────────────────────┘
Six layers of safety between an AI agent and your production infrastructure. Not because we don't trust AI — because we trust good architecture.