Simon Willison open-sources three external sandboxes for safer coding agents
Simon Willison · rss · 2026-11-18
Simon Willison, working with Prime Radiant, introduces the external sandbox pattern: the agent lives outside the sandbox while its executed code runs inside, making it safer to try local models without risking damage from mistakes or prompt injection.
Three released implementations share the same localhost API:
- external-sandbox-v86: Node.js server running Alpine Linux in WebAssembly via v86
- external-sandbox-container: Python server using Alpine Linux in an Apple Containers instance
- external-sandbox-quicksand: Python server using Microsoft's Quicksand library (bundled QEMU + Alpine)
Pi, Claude Code, or Codex can be configured to route code execution through the sandbox. Includes an optional HTTP/HTTPS allow-list proxy to control exactly which hosts agent code may reach, plus a credential replacement mechanism for authenticated API calls without exposing secrets to the agent.
More from coding & agent
- Code Mode turns 1: every major agent harness has adopted it — and kept the name — ritakozlov · 2026-11-18
- GitHub and Microsoft's MDASH pilot turns thousands of AI vuln findings into 99 verified fixes — mustafasuleyman · 2026-11-18
- OpenAI updates API dashboard with per-model usage, rate-limit and error tracking — OpenAIDevs · 2026-11-18
- 'Proof-Carrying Calls' Idea: Make Agents Certify They Don't Break Your Invariants — tianyin_xu · 2026-11-18
- Devin Platform Adds a 'Poke' Bot Template: An Engineering Agent with Personality — marvinvonhagen · 2026-11-18
- AWS cuts agent token costs 71% and latency 80% on 23MB JSON workloads with three strategies — AWS ML Blog · 2026-11-18