Skip to content
Conduit

mcp server

neuracode

The Next-Generation AI Agent Cognitive Enhancement System

Set up

Runs on your computer. Pick your agent and operating system.

This project does not publish an install method we can turn into a config. Follow the README instead.

README

synced Oct 21, 2026

🧠 NeuraCode


🌟 What is NeuraCode?

NeuraCode is a revolutionary cognitive enhancement system for AI coding agents. Unlike traditional approaches that compress and limit agent capabilities, NeuraCode enhances agent intelligence by providing:

  • 🧠 Deep Codebase Understanding - Build a complete knowledge graph of your code
  • 🔮 Predictive Context - Anticipate what agents need before they ask
  • 📚 Continuous Learning - Get smarter with every interaction
  • 🎨 Multi-Modal Understanding - Understand diagrams, architecture, and whiteboards
  • 🤝 Collaborative Reasoning - Multiple agents working together
  • 🌐 Universal Agent Support - Works with 10+ AI coding agents

🚀 Quick Start

Installation

# Install NeuraCode CLI
curl -fsSL https://raw.githubusercontent.com/kevindurant735rocket-creator/neuracode/main/scripts/install.sh | bash

# Or install via cargo
cargo install neuracode-cli

Initialize Your Project

cd your-project
neuracode init

Index Your Codebase

neuracode index

Search Your Code

neuracode search "authentication"

Predict Context for Tasks

neuracode predict "fix login bug"

Install for AI Agents

# Install for all supported agents
neuracode install --all

# Or install for a specific agent
neuracode install --agent claude-code

🎯 Core Features

1. Code Brain (代码大脑)

Deep understanding of your codebase:

# Semantic search
neuracode search "user authentication"

# Impact analysis
neuracode impact src/auth.ts

# Architecture detection
neuracode architecture

# Hotspot identification
neuracode hotspots

Capabilities:

  • ✅ Code graph with functions, classes, modules
  • ✅ Dependency analysis
  • ✅ Call chain tracing
  • ✅ Architecture pattern detection
  • ✅ Hotspot identification

2. Predict Engine (预测引擎)

Anticipate what agents need:

# Predict context for a task
neuracode predict "refactor user module"

Predicts:

  • Relevant files
  • Test files
  • Recent commits
  • Error logs
  • Similar fixes
  • Architecture context
  • Dependencies

3. Learn Engine (学习引擎)

Continuous learning from sessions:

  • ✅ Code style detection
  • ✅ Pattern recognition
  • ✅ User preference learning
  • ✅ Knowledge accumulation
  • ✅ Cross-session memory

4. Multi-Modal (多模态理解) — 🚧 Experimental

Status: framework only. neuracode understand currently detects image type from the file name and returns placeholder text; vision-model/OCR integration is on the roadmap (see docs/roadmap.md).

Understand visual content:

# Understand architecture diagrams
neuracode understand diagrams/architecture.png

# Parse flowcharts
neuracode understand diagrams/flowchart.png

# Read whiteboard photos
neuracode understand photos/whiteboard.jpg

Supports:

  • Architecture diagrams
  • Flowcharts
  • Sequence diagrams
  • Class diagrams
  • Whiteboard photos
  • Screenshots

5. Collaborative Reasoning (协作推理) — 🚧 Roadmap

Status: task-decomposition scaffolding exists; parallel execution and knowledge fusion currently return mock results. Real multi-agent orchestration is on the roadmap (see docs/roadmap.md).

Multiple agents working together:

  • ✅ Task decomposition
  • ✅ Parallel execution
  • ✅ Knowledge fusion
  • ✅ Conflict resolution
  • ✅ Collective intelligence

6. Multi-Agent Support (多Agent支持)

Works with all major AI coding agents:

AgentCore configInstall script
Claude Code✅ Supported✅ skills/install/claude_code.sh
Cursor✅ Supported✅ skills/install/cursor.sh
Codex✅ Supported✅ skills/install/codex.sh
Gemini CLI✅ Supported🚧 Roadmap
OpenCode✅ Supported🚧 Roadmap
GitHub Copilot✅ Supported🚧 Roadmap
Windsurf✅ Supported🚧 Roadmap
Cline✅ Supported🚧 Roadmap
Aider✅ Supported🚧 Roadmap
Continue✅ Supported🚧 Roadmap

📊 Performance

No benchmark numbers are claimed here. Measured figures live in docs/benchmarks.md (run cargo test -p neuracode-core for the regression suite; criterion benches under crates/neuracode-core/benches/).

CapabilityMethod
Indexingtree-sitter parse + cached file scan (see docs/benchmarks.md)
Searchkeyword + stem + fuzzy recall, no vector index yet (see docs/retrieval.md)
Language parsingRust/JS/TS/Python/Go/Java/C/C++ via tree-sitter; others file-level only
Agent support10 core configs, 3 install scripts

🏗️ Architecture

┌─────────────────────────────────────────────────────────────────┐
│                        NeuraCode                                 │
├─────────────────────────────────────────────────────────────────┤
│                                                                 │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────┐             │
│  │   Code      │  │   Predict   │  │   Learn     │             │
│  │   Brain     │  │   Engine    │  │   Engine    │             │
│  │             │  │             │  │             │             │
│  │ Code Graph  │  │ Task Class  │  │ Pattern Rec │             │
│  │ Dependencies│  │ Context Pred│  │ Preference  │             │
│  │ Architecture│  │ Pre-fetch   │  │ Knowledge   │             │
│  └─────────────┘  └─────────────┘  └─────────────┘             │
│                                                                 │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────┐             │
│  │   Multi-    │  │   Collab    │  │   Multi-    │             │
│  │   Modal     │  │   Reasoning │  │   Agent     │             │
│  │             │  │             │  │             │             │
│  │ Images      │  │ Multi-Agent │  │ Claude Code │             │
│  │ Diagrams    │  │ Knowledge   │  │ Cursor      │             │
│  │ Whiteboards │  │ Fusion      │  │ Codex       │             │
│  └─────────────┘  └─────────────┘  └─────────────┘             │
│                                                                 │
└─────────────────────────────────────────────────────────────────┘

🛠️ Development

Prerequisites

  • Rust 1.70+
  • Python 3.10+
  • Cargo

Build

# Clone the repository
git clone https://github.com/kevindurant735rocket-creator/neuracode.git
cd neuracode

# Build the project
cargo build --release

# Run tests
cargo test

# Install Python dependencies
pip install -e python/

Project Structure

neuracode/
├── crates/
│   ├── neuracode-core/       # Core Rust library
│   ├── neuracode-cli/        # CLI tool
│   └── neuracode-mcp/        # MCP server
├── python/
│   └── neuracode/            # Python AI modules
├── skills/                   # Agent skills
├── docs/                     # Documentation
└── tests/                    # Tests

📖 Documentation


🤝 Contributing

We welcome contributions! Please see our Contributing Guide for details.


📄 License

NeuraCode is licensed under the MIT License. See LICENSE for details.


🌟 Star History

Star History Chart


🙏 Acknowledgments

  • tree-sitter for AST parsing
  • CodeBERT for code embeddings
  • All the amazing open-source projects that inspired NeuraCode

↑↓ navigate openesc close