Getting started
Start with the system model: clients request and display work; backend/control-plane authority decides what is allowed and what state is authoritative.
Documentation
Public documentation for the ILAIOS system model.
This documentation overview covers stable architectural principles that can be described publicly today. API references and release-specific integration instructions will be published only when the corresponding interfaces are ready and verified.
Architecture
Understand the separation between clients, control plane, policy, execution tools, validation, and evidence.
Security model
Review permission boundaries, approval gates, least-privilege execution, and auditable state transitions.
Execution model
Prefer deterministic execution when available; use intelligent capabilities inside explicit task and tool contracts.
Agent model
Agents are bounded capabilities with defined roles, permissions, inputs, outputs, dependencies, and escalation paths.
Evidence model
Important actions should produce durable records that allow validation, review, and recovery without relying on narrative memory.