Key Features

Nova integrates agents, nodes, and protocols to power secure, composable AI execution. Nova’s architecture consists of 3 core components: the Model Context Protocol (MCP), Nova Nodes and AI Agents.

Model Context Protocol (MCP)

The MCP defines how agents interface with tools and data sources. It ensures agents can request and act upon information with verifiable integrity.

• Secure Context Windows: Enable ephemeral, encrypted sessions where agents and tools can share sensitive state.

• Tool Binding Layer: Converts external APIs or smart contracts into callable modules, abstracting infrastructure complexity.

• Audit Trail: Every interaction is signed and recorded, creating a transparent ledger of agent activities and tool responses.

---

Nova Nodes

Nova Nodes are stateless, decentralized runtimes that execute agent workloads and validate their contextual integrity.

• Validator Layer: Maintains the consistency and trustworthiness of context data.

• Execution Sandbox: Supports various LLMs within secure, containerized runtimes for task execution.

• Rewards Engine: Uses $NOVA to compensate node operators based on performance metrics such as uptime, execution accuracy, and task coverage.

---

AI Agents

AI Agents on Nova are lightweight, modular, and permission-controlled. They function as autonomous digital workers.

• Declarative Agent Definitions: Written in YAML with embedded prompts, allowing transparency and version control.

• Hybrid Execution Model: Triggers can originate from on-chain events, while actions execute off-chain.

• Access Control & Reputation: Agents and tools leverage cryptographic keys and accrue usage-based reputation.