[Bug/Regression] MCP tool result wrapped as `multimodal_text` + `parts[]` — model cannot extract structured data, chained tool call fails with fabricated ID

That’s fantastic news, thank you so much for sharing! I can confirm on our end as well — I’ve just tested and the issue is fully resolved. No more multimodal_text wrapping, tool chaining is working correctly again.

Really glad to hear it’s fixed across multiple MCP integrations, not just ours. And kudos for the methodical approach — having Claude write the compliance tests to isolate the problem at the ChatGPT integration layer is exactly the kind of rigorous debugging that helps move things forward.

Hopefully this one sticks this time!