Your LLM is a brain in a jar.
Give it a nervous system.

A biological architecture for artificial minds. Sensation, reflex, memory, and action—organized into coherent loops that turn chatbots into organisms.

Get Started View on GitHub
4 Nervous System Layers 13 Specialized Modes 5 MCP Tools
The Agentic Nervous System
🧠
CENTRAL (Brain)
Cognition & Planning
Orchestrator • Architect • Planner • Ask
Motor commands ↓ Sensory input ↑
🦾
SOMATIC (Body)
Voluntary Action
Red Phase • Green Phase • Blue Phase • Code
💭
AUTONOMIC (Subconscious)
Background Processing
Memory • Synch MCP • Index Foundry
⚡
REFLEX (Spinal Cord)
Immediate Validation
Trace MCP • Schema Validation • Contracts
The Problem

LLMs are chatbots, not agents

  • High intelligence, zero agency
  • No persistent memory between sessions
  • Tools are just sticks to poke with
  • No coordination between capabilities
The Solution

A nervous system architecture

  • Closed feedback loops of action and sensation
  • Autonomic background state management
  • Reflex validation before cognition
  • Specialized modes with clear contracts

Meet the Team

Central Layer

Planning & Architecture

  • 🔄 Orchestrator - Task coordination
  • 🏛️ Architect - System design & ADRs
  • 📋 Planner - Task maps & backlogs
  • 🔎 Deep Research - Analysis & research
Somatic Layer

TDD Development Cycle

  • 🔴 Red Phase - Write failing tests
  • 🟢 Green Phase - Implement to pass
  • 🔵 Blue Phase - Refactor & polish
  • 💻 Code - Complex implementation
Autonomic Layer

Quality & Knowledge

  • 🐛 Debug - Root cause analysis
  • 📁 Memory - Documentation
  • ❓ Ask - Clarification & research
  • 🔬 Deep Scope - Issue scoping
View all 13 agents

MCP Ecosystem

Each tool maps to a nervous system layer. Start with Central (toolless), add Somatic, then Autonomic and Reflex as needed.

🦾
Somatic Layer

OODA MCP

62 tools for computer automation. File operations, screen capture, keyboard/mouse control, batch processing.
💭
Autonomic Layer

Synch MCP

Memory persistence, agent handoff, locks, bug tracking. Cross-session context that makes agents feel continuous.
📚
Autonomic Layer

Index Foundry

RAG pipelines from documents, websites, and code. Semantic memory for document Q&A.
⚡
Reflex Layer

Trace MCP

Schema validation, contract enforcement, type-safe scaffolding. Catch bugs before runtime.
📄
Central Layer

arXiv MCP

Search and retrieve academic papers. Research-driven development for Deep Research mode.
View all tools

Progressive Enhancement

Start toolless, add capabilities as needed. Each tier preserves all contracts from the previous.

0

Toolless

Modes + contracts only. Works with any IDE.
1

Native

+ Built-in file I/O and commands
2

Somatic

+ OODA MCP (62 tools)
3

Autonomic

+ Synch + Index Foundry
4

Full System

+ Trace MCP validation

Works with your IDE

Pre-configured templates for popular AI-assisted development environments.

🦘

Roo Code

custom_modes.yaml + .roo/

⚡

Claude Code

CLAUDE.md configuration

🎯

Cursor

.cursorrules files

🤖

Copilot

copilot-instructions.md

View all templates

Ready to give your AI a nervous system?

Start with the toolless baseline and progressively add capabilities.

Get Started View on GitHub

Find this useful? Support the project!

☕ Buy Me a Coffee