langchain-ai/langchainjs langchain@1.4.6

on GitHub

Patch Changes

• #11012 4aeaa76 Thanks @christian-bromann! - feat(langchain): add when predicate to human-in-the-loop middleware

  Add an optional when callback on InterruptOnConfig so callers can
  dynamically skip interrupts for specific tool calls. The predicate receives
  a ToolCallRequest (batch afterModel context) and returns whether to
  interrupt or auto-approve, matching Python HumanInTheLoopMiddleware.

• #11080 e51478a Thanks @aolsenjazz! - feat(aws): bedrock prompt caching middleware

  Adds bedrockPromptCachingMiddleware. The interface largely matches what was previous implemented with anthropicPromptCachingMiddleware, making a best-effort attempt at utilizing prompt caching features for supported models.