# sandbox-mcp — AI Coding Tool Review & Pricing

> sandbox-mcp gives an AI agent a fresh, disposable Windows 11 machine for testing apps. Its `sbx` CLI and optional MCP adapter boot Windows Sandbox, deploy a build, drive the UI, collect screenshots and logs, and then delete the guest, so agents can verify that software runs on a clean PC rather than only on the developer's machine.

Source: https://ai.dosa.dev/tools/sandbox-mcp · Updated Nov 10, 2026 · Category: AI Testing & Quality Assurance · Published by dosa.dev

## At a glance

- **Tool**: sandbox-mcp
- **Company**: Scott Hanselman
- **Category**: [AI Testing & Quality Assurance](https://ai.dosa.dev/category/qa-testing)
- **Pricing model**: Open Source
- **Official website**: https://shanselman.github.io/sandbox-mcp/

## Key features

- `sbx` CLI for one-shot runs, installer tests and long-lived sessions
- Optional MCP server exposing 27 tools with inline screenshots
- UI Automation clicks and interactive-desktop screenshots tied to a window
- Network-off guests to catch missing runtime dependencies such as VC++ redistributables
- Incremental warm deployment with restart and rollback
- Bounded desktop recording with HTML playback and optional MP4
- Agent-led setup guide with checksum verification and consent before admin changes

## Pricing

**Model**: Open Source

sandbox-mcp is free and open-source under the MIT license. It relies on Windows Sandbox, which is built into Windows 11 Pro, Enterprise and Education (24H2 or later); Windows Home and Windows Server are not supported.

## Who it is for

- **Best for**: Windows desktop app developers who want Copilot, Claude, Cursor or other MCP-capable agents to test builds and installers on a clean Windows machine.
- **Not ideal for**: macOS or Linux users, Windows Home editions, or teams that need cross-platform or cloud-hosted test machines.

## Verdict

A practical, well-documented bridge between coding agents and Windows Sandbox for clean-machine app testing; Windows-only and very new.

## Recent updates

v1.2.0, released on November 7, 2026, added desktop recording, encrypted guest transport, warm deployment, MSIX/Appx handling and an agent-led setup guide, one day after v1.1.0.

## Frequently asked questions

### Is sandbox-mcp free?

Yes - sandbox-mcp is Open Source. sandbox-mcp is free and open-source under the MIT license. It relies on Windows Sandbox, which is built into Windows 11 Pro, Enterprise and Education (24H2 or later); Windows Home and Windows Server are not supported.

### Is sandbox-mcp open source?

Yes - sandbox-mcp is open source. sandbox-mcp is free and open-source under the MIT license. It relies on Windows Sandbox, which is built into Windows 11 Pro, Enterprise and Education (24H2 or later); Windows Home and Windows Server are not supported.

### Who is sandbox-mcp best for?

Windows desktop app developers who want Copilot, Claude, Cursor or other MCP-capable agents to test builds and installers on a clean Windows machine.

### Who is sandbox-mcp not ideal for?

macOS or Linux users, Windows Home editions, or teams that need cross-platform or cloud-hosted test machines.

### What are the best sandbox-mcp alternatives?

The closest sandbox-mcp alternatives on ai.dosa.dev are GitAuto, AgentDiff, Agent QA - all listed under AI Testing & Quality Assurance.

### Who makes sandbox-mcp?

sandbox-mcp is developed by Scott Hanselman. It is listed in the AI Testing & Quality Assurance category on ai.dosa.dev.

## Top alternatives

- [GitAuto](https://ai.dosa.dev/tools/gitauto) (Unknown) — Development teams with legacy codebases or high test coverage requirements who need to automate repetitive unit testing and maintenance tasks.
- [AgentDiff](https://ai.dosa.dev/tools/agentdiff) (Freemium) — Engineers building multi-turn, tool-calling agents who need to verify specific execution paths and maintain provenance in production environments.
- [Agent QA](https://ai.dosa.dev/tools/agent-qa) (Open Source) — Engineering teams that want to own their testing stack, maintain test and memory files within their own repository, and leverage AI agents for automated QA workflows.
- [e2e](https://ai.dosa.dev/tools/e2e) (Open Source) — Teams looking to combine the flexibility of agent-driven testing with the precision of scripted, locator-based E2E tests, particularly for projects that frequently evolve or require high-speed CI feedback.
- [iFixAi](https://ai.dosa.dev/tools/ifixai) (Freemium) — Engineering teams shipping AI agents that handle sensitive tasks like financial transactions, database edits, or internal tool usage.
- [ARTEMIS](https://ai.dosa.dev/tools/artemis) (Open Source) — Mobile developers and QA engineers who need an autonomous AI agent to execute end-to-end testing, bug reproduction, or exploratory stability checks on Android devices.
- Full alternatives guide: https://ai.dosa.dev/tools/sandbox-mcp/alternatives