Detailed Introduction
Hopx is a cloud runtime designed for AI agents, providing millisecond startup isolated microVMs and multi-language SDKs to run LLM-driven agents in secure, controlled execution environments. Hopx supports persistent sessions, filesystem access, background processes, and real-time output streaming, enabling stable execution of long-running agents and complex automated workflows.
Main Features
- Millisecond cold starts for fast interactive workloads.
- VM-level isolation using Firecracker for stronger security boundaries than containers.
- Multi-language SDKs (Python, JavaScript/TypeScript, Go, etc.) with MCP integration.
- Persistent sessions, real-time streaming logs, and runtime monitoring.
Use Cases
Suitable for running untrusted code in isolated environments: LLM-driven agents, interactive notebooks, long-running jobs, desktop automation, and as an MCP backend for assistants and tools. See the official documentation for integration guides.
Technical Characteristics
Hopx leverages microVMs and snapshotting to combine strong isolation with fast startup times. Templates and SDKs provide a unified runtime API, while Firecracker and snapshot-based workflows enable safe, fast, and scalable execution of agents and workflows in production environments.