Source Intelligence

DisclaimerUnofficial, and not affiliated with Anthropic. Nearly all of this is read straight out of what ships: npm bundles, captured prompts, published docs. Anthropic's own notes go in verbatim, marked as theirs. The rest is my reading, and every entry carries the strings behind it. If one looks wrong, vote it down and say why.

All of v2.0.34 Home All releases olderv2.0.33 v2.0.35newer
Claude Code v2.0.34

Sandbox Override Configuration

What: New settings UI to control whether commands can fall back to running outside the sandbox when sandbox restrictions prevent execution.

How to use:

# Access via /sandbox command in interactive mode
# Choose from:
# 1. "Allow unsandboxed fallback" - Commands retry outside sandbox if blocked
# 2. "Strict sandbox mode" - All commands must run in sandbox or be excluded

Details:

  • Separate from the main sandbox enable/disable setting
  • Can be locked by policy settings from higher-priority configuration sources
  • Status displayed in /sandbox command output (e.g., "sandbox enabled, fallback allowed")
  • New API methods: areUnsandboxedCommandsAllowed(), areSandboxSettingsLockedByPolicy(), setSandboxSettings()
  • Documentation link: docs.claude.com/en/docs/claude-code/sandboxing#configure-sandboxing
  • Evidence: Component bn2() at line 465885, policy check Fs2() at line 472000 in pretty-v2.0.34.js

See this entry in the whole of v2.0.34 →