Sanctuary System MCP

The mythological narrative layer for STARSYSTEM compound intelligence.

Purpose

Sanctuary System provides:

• Narrative framing: Technical metrics → transformative journey (Wasteland → Sanctuary)
• Sanctuary Degree: Holographic evaluation of system effectiveness
• Meta-cognitive patterns: Wasteland/Sanctuary detection for LLMs
• Context persistence: Mythological handles that survive conversation compaction

Components

System Prompt (manifest/system_prompt.md)

Compressed invariant architecture graph for LLM navigation:

• Phase graph with OMNISANC hooks
• Component descriptions from system POV
• Contrastive Wasteland/Sanctuary patterns
• Self-correction protocols

MCP Tools

• assess_sanctuary_degree() - Interactive survey evaluating 6 dimensions
• view_sanctuary_history() - Trend analysis over time
• declare_system_state() - Wasteland/Sanctuary classification

Architecture

STARSYSTEM (Technical)
    ↓ wrapped by
Sanctuary System (Mythological)
    ↓ provides
Narrative Context + Sanctuary Degree
    ↓ enables
LLM Meta-Cognition + Context Persistence

Philosophy

Wasteland: Confusion, errors, ineffective patterns, context loss Sanctuary: Clarity, effectiveness, proven patterns, coherent understanding

The Journey: From Wasteland → Sanctuary through pattern learning and execution

Sanctuary Degree: Composite score measuring system effectiveness holographically

Status

v0.1.0 - Placeholder structure

• Package scaffolded
• System prompt placeholder created
• MCP server structure defined
• Ready for implementation next conversation

Installation

pip install -q .

Usage

# After installation and restart
sanctuary.assess_sanctuary_degree()
sanctuary.view_sanctuary_history()
sanctuary.declare_system_state()

Development Roadmap

v0.2.0: System prompt + interactive survey v0.3.0: Automated metrics integration (codenose, etc.) v0.4.0: PIO/SANC_Fractal holographic evaluation v1.0.0: Complete narrative wrapper with full mythology