Source Intelligence
Follow Discord

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 v1.0.127 Home All releases ←olderv1.0.126 v1.0.128newer→
Claude Code v1.0.127

Enhanced Sandbox Command Filtering

What: Added support for substring-based command filtering in the sandbox configuration through the new tengu_sandbox_disabled_commands feature gate.

How it works:

The sandbox now supports two methods for disabling commands:

  1. Exact command matching (existing unsandboxedCommands in settings)
  2. Substring matching (new tengu_sandbox_disabled_commands feature flag)

Details:

  • Substring matching allows blocking commands that contain specific patterns anywhere in the command string
  • The new tengu_sandbox_disabled_commands feature flag provides commands and substrings arrays
  • Both methods are combined when evaluating whether a command should bypass sandboxing
  • Evidence: Qg2() function at line 379010 in v1.0.127 (replaces qu2() at line 383844 in v1.0.126)

See this entry in the whole of v1.0.127 →