Add multimodal workspace read support

[threepointone]

Summary

• Add multimodal-aware Think workspace reads for images and PDFs while preserving line-numbered text reads.
• Keep persisted read tool outputs compact and use AI SDK toModelOutput to rehydrate image/file bytes for model input, including replayed history.
• Cover image/PDF/binary read behavior, replay conversion, docs, and release notes.

Test plan

• npm run test --workspace @cloudflare/think -- assistant-tools.test.ts think-session.test.ts
• npm run build --workspace @cloudflare/think
• npm run check

Made with Cursor

[changeset-bot]

🦋 Changeset detected

Latest commit: 469c7d3

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
@cloudflare/think	Patch

Not sure what this means? Click here to learn what changesets are.

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