🏛️ 6-Tier Hybrid Architecture
Complete process isolation between Core LLM (Port 3000), ML Engine (Port 8000), MCP Server, Dashboard, Policy Engine, and Signer Vault.
Autonomous on-chain execution without sacrificing self-custody. Powered by an Isolated Signer Vault, OS-native credential storage, and an On-Chain Base Sepolia Kill-Switch.
Eliminates vulnerable "Master Passwords" and plaintext .env files. Private keys are encrypted and stored directly within your operating system's trusted Keyring (GNOME Secret Service, macOS Keychain, or Windows Credential Manager).
Empower your agent with OS-level System Automation. Run shell scripts and manage local files dynamically while retaining explicit power to revoke privileges at any time.
Trust but verify. All destructive or financially sensitive transactions are paused and wait for your explicit cryptographic signature before broadcasting.
A sleek, readable dashboard built for professional Web3 execution. Manage multiple isolated chat sessions seamlessly across various AI models (Gemini, Claude, DeepSeek, etc.) in a unified workspace.
Automatically routes trades through MEV-protected aggregators and enforces strict dual-layered slippage hard-limits to block front-running and sandwich attacks.
Nyxora quietly extracts and learns your Web3 habits in the background. The asynchronous Nyx Daemon continuously audits your chat history to extract your persona and save it to episodic.db, drastically minimizing repetitive user instructions.
Extend Nyxora's capabilities effortlessly. Drop any custom community skill into your local ~/.nyxora/skills/ directory, or command Nyxora to autonomously code and synthesize a new skill for itself!
Turn Nyxora into a backend powerhouse. Seamlessly connect its sandboxed Web3 execution environment to external AI tools like Claude Desktop or Cursor via the Model Context Protocol.
Stop letting assets sit idle. Command Nyxora to autonomously supply liquidity to Aave V3 or route LP tokens into Beefy Finance auto-compounding vaults.