DeepThinking AI

Tag

sdk

Covered in AI Engineering, where the background and the sources for this subject live.

AI Engineering

How do you stop Claude Code from expanding untrusted prompt text?

Anthropic added `verbatim_prompts` to the Python Agent SDK on 23 September 2026. It marks each user message `client_composed: true`, which Claude Code 2.1.248 and later delivers as written, so `@/path` tokens do not expand into file contents and slash commands do not dispatch. For apps that inline untrusted text, this closes a pre-permission local-read gap.

4 min read