---
title: "Secure Sandbox — SDK for AI Sandbox Security | Vercel AI SDK, Pydantic AI, LangChain, E2B, Daytona, Cloudflare Containers, Blaxel, Sprites, Modal, Runloop & exe.dev"
description: "TypeScript and Python SDK that wraps any hosted AI sandbox with kernel-level security. Works with Vercel AI SDK, Pydantic AI, LangChain, and any AI framework. Supports Vercel Sandbox, E2B Sandbox, Daytona Sandbox, Cloudflare Containers, Blaxel Sandbox, Sprites (Fly.io), Modal, Runloop, exe.dev, Freestyle, and Tensorlake."
doc_version: "1.0"
last_updated: "2026-06-05"
canonical: "https://www.agentsh.org/secure-sandbox/"
---

# Secure Sandbox

## Overview

Secure Sandbox is the TypeScript and Python SDK layer for applying agentsh runtime enforcement to hosted AI sandboxes. It wraps sandbox providers with policy enforcement for package installs, file access, commands, network egress, and secret handling.

The SDK keeps provider choice reversible. If one workload starts in E2B, another runs in Vercel, and a third needs Modal or Runloop for GPU or devbox behavior, teams can keep the same policy vocabulary around all of them. File rules, network allowlists, package checks, command restrictions, and secret redaction remain the operational contract even when the underlying sandbox changes.

## Supported Platforms

- Vercel Sandbox
- E2B Sandbox
- Daytona Sandbox
- Cloudflare Containers
- Blaxel Sandbox
- Sprites/Fly.io
- Modal
- Runloop
- exe.dev
- Freestyle
- Tensorlake

## Related Documentation

- [Secure Sandbox SDK reference](https://www.agentsh.org/docs/secure-sandbox/)
- [Execution-layer security overview](https://www.agentsh.org/execution-layer-security/)
- [Glossary](https://www.agentsh.org/docs/glossary/)

## Sitemap

- [Canonical HTML](https://www.agentsh.org/secure-sandbox/)
- [Site map](https://www.agentsh.org/sitemap.md)
- [Full documentation](https://www.agentsh.org/llms-full.md)