Meet Anthropic’s sandbox-runtime: OS-Level Sandboxing for AI Agents, No Containers
MarkTechPost Asif Razzaq
Anthropic released an open-source sandbox-runtime that wraps processes with OS-level filesystem and network restrictions instead of using containers. The npm package is @anthropic-ai/sandbox-runtime v0.0.81 with about 11.4 MB unpacked size, and Anthropic reports an internal 84% reduction in Claude Code permission prompts. Developers can run Claude Code and their own agents or MCP servers inside it by prefixing commands with srt (or configuring rules via ~//.srt-settings.json), but Windows support remains alpha and security depends on how broad the allowlists are.
Why it matters
Anthropic sandbox-runtime (srt) is the open-source, OS-level sandbox behind Claude Code. It restricts file and network access for agents, MCP servers and CLI tools with Seatbelt on macOS and bubblewrap on Linux, no containers needed. The post Meet Anthropic’s sandbox-runtime: OS-Level Sandboxing for AI Agents, No Containers appeared first on MarkTechPost.
