Hi everyone,

I'm hitting a breaking issue with an ODC agentic app when trying to send a PDF file as Binary Data to an Anthropic/Claude model using the native platform actions.

This was working perfectly before. Absolutely zero changes were made to our ODC code or workflows, and we are feeding it the exact same PDFs as before.

What's happening:

The workflow runs a standard server action , passes the binary, and immediately throws a BadRequest error:

"messages.0.content.2.image.source.base64.media_type: Input should be 'image/jpeg', 'image/png', 'image/gif' or 'image/webp'"

The constraint:

It looks like Anthropic recently tightened their API validation on their backend. ODC's internal mapping layer seems to be serializing our PDF binary into an image content block (messages.0.content.2.image...), which Anthropic now strictly rejects because the MIME type doesn't match an image.

The real problem is that in ODC, we pass the identifier of a system-managed Static Entity record at runtime to handle the format. This Static Entity completely lacks a record that maps correctly to Anthropic's native document block type. Because everything is abstracted behind these Static Entity identifiers, there is literally no way for us to manually override or change the underlying JSON request payload to fix the format.

Is anyone else running into this right now? Is there an official platform patch coming to update these Static Entity records for Claude's newer document API schemas?

Thanks!