SanctumOS Components

SanctumOS is built on a neuro-inspired architecture that maps software components to cognitive functions. This approach ensures intuitive behavior and predictable system operation.

Core Cognitive Components

🧠 Thalamus

Input Preprocessing Hub

• Processes raw inputs through multiple phases
• Provides immediate pass-through for reflex responses
• Performs cleanup, re-segmentation, and semantic annotation
• Acts as the first stage of information processing

🎯 Basal Ganglia

Task Orchestrator

• Selects and manages long-running sub-agent tasks
• Enforces success/fail loop structures
• Provides action gating and habit engine functionality
• Learns from outcomes to improve future task selection

💭 Dream Agent

Memory Consolidation System

• Nightly batch processing of daily conversations
• Three coordinated neuro-modules:
• Hippocampus: Knowledge extraction and summarization
• Neocortex: Knowledge integration and storage
• Glymphatic System: Storage optimization and pruning
• Maintains efficient long-term memory

Communication Components

📡 Broca-2

Message Processing Middleware

• CLI-first, plugin-based message processor
• Translates agent decisions into platform-specific messages
• Supports multiple communication endpoints
• Provides message queue with echo, listen, and live modes

🔌 Sanctum Letta MCP

Model Context Protocol Server

• Plugin-based MCP server for AI tool integration
• Provides seamless integration between AI clients and external tools
• Supports auto-discovery and health monitoring
• Enables standardized tool communication

System Architecture

Component Interaction

User Input → Thalamus → Cerebellum → Conscious Mind
                ↓           ↓            ↓
            Memory ← Dream Agent ← Basal Ganglia
                ↓
            Broca-2 → External Platforms

User Input → Thalamus → Cerebellum → Conscious Mind ↓ ↓ ↓ Memory ← Dream Agent ← Basal Ganglia ↓ Broca-2 → External Platforms

User Input → Thalamus → Cerebellum → Conscious Mind
                ↓           ↓            ↓
            Memory ← Dream Agent ← Basal Ganglia
                ↓
            Broca-2 → External Platforms

Data Flow

1. Input Processing: Thalamus preprocesses raw inputs
2. Filtering: Cerebellum prioritizes and filters information
3. Reasoning: Conscious Mind performs complex reasoning
4. Task Management: Basal Ganglia orchestrates sub-agent tasks
5. Memory Consolidation: Dream Agent maintains long-term memory
6. Communication: Broca-2 handles external communication

Design Principles

Neuro-Inspired Architecture

Each component is designed to mirror human cognitive functions:

• Thalamus: Like the brain's sensory relay station
• Basal Ganglia: Like the brain's action selection system
• Dream Agent: Like the brain's memory consolidation during sleep
• Broca: Like the brain's language production center

Modular Design

• Independent Components: Each component can operate independently
• Clear Interfaces: Well-defined communication protocols
• Extensible Architecture: Easy to add new components
• Plugin System: Support for custom extensions

Privacy and Control

• Self-Hosted: All components run on your hardware
• Transparent: Full visibility into system operations
• Auditable: Complete logging and provenance tracking
• User-Controlled: You maintain full control over your data

Related Documentation

• Architecture Overview: Complete system architecture
• Naming Rubric: Consistent terminology guidelines
• Module Documentation: Detailed module information

---

SanctumOS Components - A neuro-inspired approach to building intelligent, self-hosted AI systems.