add streamText to aisdk llm client

[tkattkat]

why

• Currently we have no way to utilize streamtext, only generateObject, and generateText

what changed

• Streamtext can now be utilized through createChatCompletion by passing stream:true

test plan

• tested locally

[changeset-bot]

⚠️ No Changeset found

Latest commit: a8fdc92

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[greptile-apps]

Greptile Summary

This review covers only the changes made since the last review (commit 111ec3b), not the entire PR.

The most recent changes focus on implementing streaming text functionality in the AI SDK LLM client. The developer added three new optional properties to the ChatCompletionOptions interface: aiSDKTools (of type ToolSet), maxSteps (number), and stream (boolean). In the AISdkClient implementation, a new conditional branch was added that checks for options.stream and uses the streamText function from the AI SDK when streaming is enabled.

When stream: true is passed, the client now calls streamText with the formatted messages, temperature, AI SDK tools, and max steps parameters, then returns the stream directly cast as T. The implementation also updates tool parameter references from options.tools to options.aiSDKTools throughout the client to maintain consistency with the interface changes.

This change enables real-time text generation through the existing createChatCompletion interface, allowing users to receive partial responses as they're generated rather than waiting for complete responses. The streaming functionality bypasses the existing caching mechanism entirely, which is necessary since streams cannot be cached or processed synchronously like regular responses.

Confidence score: 3/5

• This PR introduces useful streaming functionality but has some implementation concerns that could cause issues in production
• Score reflects the type safety concerns with the stream casting and the fact that streaming bypasses caching mechanisms
• Pay close attention to the type casting in the streaming path and ensure consumers can handle the stream response appropriately

_{2 files reviewed, 1 comment}

_{Edit Code Review Bot Settings | Greptile}