Vercel has updated its AI SDK, with releases for @ai-sdk/valibot@3.0.30, @ai-sdk/voyage@2.0.30, @ai-sdk/workflow@2.0.9, and @ai-sdk/xai@4.0.44. These updates include the introduction of new harness adapters and enhancements to message streaming and batch completion functionalities.
Key Points
- The
@ai-sdk/harness-cursor@1.0.0package implements a Cursor harness adapter. - The
@ai-sdk/harness-fx@1.0.0package implements an FX harness adapter. - The
ai@7.0.79package preserves active text and reasoning parts when another merged UI message stream finishes a step. - Workflow stream normalization has been aligned with explicit part end chunks.
- Batch completion webhooks have been added as an experimental feature.
- The
experimental_startTextBatchfunction now accepts a webhook.
Context
According to the Vercel AI SDK Changelog, these updates were processed via the Changesets release GitHub action. This mechanism allows for automated package publishing to npm upon merging the relevant pull request. The updates also include dependency changes for @ai-sdk/harness-acp@1.0.25, @ai-sdk/harness@1.0.87, and @ai-sdk/provider-utils@5.0.30.
Why It Matters
These updates provide developers using the Vercel AI SDK with new tools for integrating and managing AI model interactions, particularly through the new harness adapters and the experimental batch completion webhooks. The improvements to UI message stream handling aim to enhance the reliability and consistency of AI-driven workflows.
What To Do
- Note the new Cursor and FX harness adapters for potential integration.
- Investigate the
experimental_startTextBatchfunction and its webhook capabilities for batch processing. - Review the updated dependency versions for
@ai-sdk/harness-acp,@ai-sdk/harness, and@ai-sdk/provider-utilsin your projects. - Observe how the changes to UI message stream normalization affect existing implementations.
