Architecture

This document describes the structural organisation of Daemon Core. Not implementation detail. Architectural rationale.

This layer cannot be retrofitted by tools above it. Governance must be foundational.

Conceptual Model

Each agent receives three structural inputs before operation:

• A rulebook defining behavioral constraints.
• A scope boundary delimiting operational range.
• A context focus specifying current workload.

These map to three architectural layers: structure, boundaries, coordination.

Layer 1: Structure

The structural layer governs agent initialisation:

• Core rules applied universally.
• Global preferences and configurations.
• Identity and capability declarations.

This layer ensures behavioral consistency across sessions and restarts.

Layer 2: Boundaries

The boundary layer controls safety and scope:

• Operational range constraints.
• Permission matrices by environment trust.
• Human approval requirements for critical operations.

Boundaries are structural, not advisory. Agents cannot exceed defined limits.

Layer 3: Coordination

The coordination layer manages multi-agent operation:

• Context handoff protocols between agents.
• Conflict prevention through scope isolation.
• Audit trails for operational history.

Coordination is structural. Agents operate as components of a unified system, not isolated instances.

Extension Model

Daemon Core supports extension through defined interfaces:

• Custom role profiles for domain-specific agents.
• Environment-specific boundary configurations.
• Integration hooks for external tooling.

Extensions operate within core constraints. They cannot override fundamental safety or coordination mechanisms.