> For the complete documentation index, see [llms.txt](https://nuvex-ai.gitbook.io/nuvex-ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://nuvex-ai.gitbook.io/nuvex-ai/developer-guide/architecture-overview.md).

# Architecture Overview

**Built on the MCP (Model-Controller-Protocol) Framework**

Nuvex is built on the advanced **Model-Controller-Protocol (MCP)** architecture — a modular design pattern that enables the development of high-performance, deeply composable intelligent agent systems.

This architecture supports everything from agent orchestration to real-time protocol communication, delivering scalability, interoperability, and developer flexibility across the stack.

***

### **System Architecture: Three-Layer Design**

#### 1. Web3 Agent Store Layer

This is the application-facing layer where users discover, deploy, and utilize specialized agents — each designed for specific high-value workflows.

**Agent Categories Include:**

* **Analysis Agent** – For data analytics and insight generation
* **Trade Agent** – Executes trading strategies across supported protocols
* **Media Agent** – Automates content creation and media engagement
* **Research Agent** – Performs deep research and multi-source knowledge extraction
* **Custom Agents** – Extensible and programmable for domain-specific needs

***

#### 2. Nuvex Agent Protocol Layer

The core intelligence layer. Here, agents plan, reason, coordinate, and act through tightly integrated services and control interfaces.

**Service Components:**

* **MCP Service** – Implements the foundational Model-Controller-Protocol architecture
* **SSE (Server-Sent Events)** – Enables real-time, persistent communication
* **CMD Interface** – Allows agent-level command control
* **HTTP Service** – Provides RESTful endpoints for third-party integrations
* **OpenAPIs** – Standardized interfaces for tool interoperability
* **SDKs** – Client libraries for integrating Nuvex agents into any development environment

***

**Agent Orchestration:**

* **Planner Agent** – Acts as a coordinator, decomposing complex tasks into modular subtasks
* **Task Agents** – Specialized executors handling defined subtasks
* **Integrated Tooling:**
  * **CodeAct** – For code generation, testing, and execution
  * **Browser** – Supports real-time information retrieval
  * **Initial Tools** – Built-in utilities for immediate deployment
  * **Search** – Cross-platform search agent
  * **Custom Tools** – Developer-defined or domain-specific extensions

***

**Client Integration Layer:**

* **Tools Center** – Registry and discovery layer for agent-compatible tools
* **Authorization Service** – Security controls for API access and agent action permissions
* **MCP Client Interfaces** – Endpoints for mobile, web, and desktop environments

***

#### 3. Chain Foundation Layer

The infrastructure layer — connecting intelligent agents to the decentralized economy and cross-platform services.

**Integrations Include:**

* **Multi-Chain Support** – Compatible with BASE, BTC, ETH, BNB, SOL, APT, SUI, and more
* **Social Platforms** – Integrates with networks like X (Twitter) and Telegram
* **DeFi Infrastructure** – Native support for CEX/DEX operations
* **Third-Party Systems** – Modular extensions for external platform interoperability

***

### **Core System Components**

#### Agent System Architecture

* **Agent Core** – Central runtime for logic processing, tool coordination, and decision modeling
* **Memory System** – Manages short-term and long-term memory states, including historical context referencing
* **Tool Manager** – Handles tool registration, validation, and execution permissions
* **Prompt Engine** – Framework for building advanced task templates and prompt chaining
* **LLM Connector** – Supports model routing, fallback strategies, and multi-model logic

***

#### Tool Support Framework

Nuvex supports a range of tool types with standardized integration paths:

* **OpenAPI Tools** – Auto-integrates third-party REST APIs via schema import
* **Blockchain Tools** – Handles wallet access, contract interaction, and multi-chain state retrieval
* **Data Analysis Tools** – For large-scale parsing, statistical modeling, and prediction
* **Custom Tools** – Developers can build and register proprietary modules via the SDK

***

### **Security Architecture**

Nuvex is secured by a layered protection model designed for agent autonomy at scale:

* **Permission Control** – Role-based access and API-level restrictions
* **Resource Limits** – Automated monitoring of memory, compute, and execution time
* **Audit Logging** – Full operation-level logging and traceability
* **Vulnerability Protection** – Hardening strategies against common agent-related exploits