Skip to content

Explanation

The explanation section provides understanding-oriented documentation that clarifies concepts and design decisions in cyberian.

What are Explanations?

Explanations help you understand:

  • How things work under the hood
  • Why things are designed the way they are
  • Connections between different components
  • Background and context for features

If you need step-by-step instructions, see Tutorials or How-To Guides instead.

Available Topics

Architecture

Understand cyberian's system architecture:

  • How cyberian wraps agentapi
  • HTTP communication model
  • Client-server relationship
  • Process management
  • Server farms architecture

Workflow Model

Deep dive into workflow execution:

  • Recursive task tree structure
  • Depth-first traversal algorithm
  • Completion detection mechanism
  • Loop execution model
  • Template rendering pipeline
  • Success criteria validation

Agent Lifecycle

Understand agent state management:

  • Server lifecycle (start, run, stop)
  • Agent memory and context
  • reuse vs refresh modes
  • When to use each mode
  • State isolation patterns

Design Decisions

Learn why cyberian works the way it does:

  • Why YAML for workflows?
  • Why Jinja2 for templates?
  • Why synchronous polling?
  • Why recursive task trees?
  • Trade-offs and alternatives

Reading Guide

These explanations are best read:

  1. After completing the tutorials
  2. When you're curious about how things work
  3. Before making architectural decisions in your workflows
  4. To understand error messages and behavior

See Also