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.
Tool errors can be rendered using what the tool was asked to do
Under the hood
Useful2Signal2
Elsewhere
Tool errors can now be rendered with knowledge of the original request, not just the error text.
What's wrong with this entry?
Anonymous. No account, no email.
What
The transcript now hands the tool's input and the list of active agents to the tool-error renderer, so an error can be displayed with knowledge of the original request rather than only its error text.
Evidence
renderToolUseErrorMessage
Strings lifted out of the shipped bundle, so the claim above can be checked against them.