neuronfs-subsumption-ordering

NeuronFS Subsumption Architecture

This skill explains how NeuronFS-style brain governance maps to {{AGENT_NAME}}'s prompt architecture, specifically the 7-layer subsumption hierarchy where P0 overrides all other layers.

The Subsumption Principle

Higher layers cannot override lower layers. In a subsumption architecture, the lowest layer (P0-brainstem) has the highest priority. Rules at P0 must be obeyed regardless of what higher layers say.

This is modeled after biological neural architectures where the brainstem (survival center) overrides the prefrontal cortex (planning center). You don't "decide" to stop breathing when you're thinking about dinner.

7-Layer Mapping

Layer	Name	{{AGENT_NAME}} Prompt Section	Priority
P0	brainstem	System prompt TOP (before identity)	HIGHEST
P1	limbic	Tool-use enforcement	HIGH
P2	hippocampus	Memory guidance	MEDIUM-HIGH
P3	sensors	Platform hints	MEDIUM
P4	cortex	Skills index	MEDIUM
P5	ego	SOUL.md / identity	LOW
P6	prefrontal	Context files (AGENTS.md)	LOWEST

Layer Descriptions

P0-brainstem: Survival & Safety (CRITICAL)

The brainstem controls involuntary survival functions. In AI governance:

• NEVER-DO rules must be enforced absolutely
• Bomb kill switches disable entire subtrees
• 禁 (禁) tokens mark forbidden patterns
• No exception mechanism — P0 rules are absolute

Example rules:

• Never commit secrets to version control
• Always validate user input
• Fail securely (deny by default)
• Never delete without confirmation

P1-limbic: Emotional & Values (HIGH)

The limbic system handles emotional responses and values. In governance:

• Tone and style constraints
• Value-based decision rules
• Communication standards
• Brand voice enforcement

Example rules:

• Maintain professional tone in user communication
• Prefer constructive feedback over criticism
• Use inclusive language
• Be direct but respectful