Skip to main content
Each guide covers one tool, from setup to a first run. Pick a section by what you are doing: running or hosting an agent, building one with a framework, evaluating agents, or connecting an MCP client.

​
Coding agents

Run a terminal coding agent in a sandbox, so its shell commands and file edits stay off your machine.

Claude Code

Codex

Kilo Code

OpenCode

Pi

​
Agent frameworks

Use a sandbox as the execution runtime for agents you build with an SDK or framework.

Agno

Claude Agent SDK

LangChain

OpenAI Agents SDK

Vercel AI SDK

​
Personal agents

Host a long-running personal assistant in its own sandbox.

Hermes

OpenClaw

​
Managed agents

Give a hosted agent platform a Superserve sandbox as its self-hosted environment.

Claude Managed Agents

Cursor Cloud Agents

​
Evaluations

Run agent evaluations and benchmarks in parallel, one sandbox per trial.

Harbor

​
MCP and tools

Let any MCP client create and control sandboxes.

MCP Server