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 v2.1.246 Home All releases olderv2.1.245 v2.1.247newer
Claude Code v2.1.246

Unverifiable permission answers now say so and confirm nothing ran

You'll notice
Useful3 Signal3
Permissions

An unverifiable permission answer now tells the model nothing ran, and the call can be retried.

What

When a permission prompt sent to another machine comes back with an answer that cannot be confirmed as yours, the model is now told "The approval for this call could not be verified as the user's own answer" and that nothing was executed, instead of a generic failure. This case is retryable.

Details
  • A companion message covers a machine you have paired whose Claude Code has never connected.
  • No flag guards this path.
Evidence

The approval for this call could not be verified as the user's own answer

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

See this entry in the whole of v2.1.246 →