Isorun
$19.36 a month*
Home / AI agent sandboxes / Anthropic sandbox-runtime
Open-source CLI and library that wraps any process in OS-level filesystem and network restrictions, no container needed.
Verdict (2026-11-13): Anthropic sandbox-runtime scores 46/100, #60 of 63 AI agent sandboxes. Strong on free tier, docs and SDKs. Weak on isolation, persistence, limits. Best for: Limiting what a local coding agent, MCP server or shell command can read, write and contact.
Skip it if: Shares the host kernel; weaker than VM-based isolation for hostile code.
Prices checked 2026-11-15.
sandbox-runtime (srt) restricts what a process and its children can read, write and reach on the network, using the operating system's own sandboxing: Seatbelt on macOS, bubblewrap on Linux and a dedicated user account with a Windows Filtering Platform fence on Windows. Network traffic goes through local proxies that enforce domain allow and deny lists. It was built for Claude Code and is published as an npm package that works as a CLI or a library. It is a research preview under Apache-2.0.
46/100 in AI agent sandboxes, rank 60 of 63. Parts with no data are left out and the rest is rescaled.
Full Anthropic sandbox-runtime pricing: plans and costs at four workloads →
$19.36 a month*
$11.73 a month*
$10.14 a month*
$19.00 a month*
$17.95 a month*
$28.44 a month*
$20.00 a month*
$20.00 a month*
Anthropic sandbox-runtime is not in the public benchmarks we track yet. See all benchmarks.
| Anthropic sandbox-runtime | |
|---|---|
| Category | agent sandbox |
| Our score | 46/100 |
| Isolation | OS-level process sandbox (sandbox-exec/Seatbelt on macOS, bubblewrap on Linux, restricted user account on Windows) with proxy-based network filtering |
| Billing | Not applicable: free software run on your own machine |
| Free tier | Free open-source software under the Apache-2.0 license. |
| Network controls | Yes |
| Open source | Yes (Apache-2.0) |
| Self-host | Yes |
| SDKs and interfaces | CLI, TypeScript/JavaScript library |
| Regions | Runs locally on macOS, Linux or Windows. |
| Facts checked | 2026-11-13 |
No reviews yet. Have you used Anthropic sandbox-runtime? Your review helps other developers choose.
Anthropic sandbox-runtime scores 46/100, #60 of 63 AI agent sandboxes. Strong on free tier, docs and SDKs. Weak on isolation, persistence, limits. Best for: Limiting what a local coding agent, MCP server or shell command can read, write and contact. Facts checked 2026-11-13.
Free open-source software under the Apache-2.0 license.
OS-level process sandbox (sandbox-exec/Seatbelt on macOS, bubblewrap on Linux, restricted user account on Windows) with proxy-based network filtering.
In AI agent sandboxes, the top-rated are Isorun (78/100), Northflank (78/100), Koyeb Sandboxes (76/100). Anthropic sandbox-runtime is #60 of 63.
A lightweight sandboxing tool for enforcing filesystem and network restrictions on arbitrary processes at the OS level, without requiring a container.
srt uses native OS sandboxing primitives (sandbox-exec on macOS, bubblewrap on Linux) and proxy-based network filtering.
By default, all network access is denied. You must explicitly allow domains.
The Sandbox Runtime is a research preview developed for Claude Code to enable safer AI agents.
npm install -g @anthropic-ai/sandbox-runtime
v0.0.83