Support audio input

[fjord-oai]

Summary

Adds audio as a structured tool-output content item, gated by model input modalities.

• Adds InputModality::Audio and input_audio function-call output items.
• Converts MCP audio content blocks to Responses-compatible structured output for audio-capable models.
• Fails MCP/code-mode audio output clearly when the selected model does not advertise audio input.
• Adds a strict code-mode audio(...) helper.
• Regenerates app-server protocol schemas so clients can round-trip the new content item shape.

[damian-openai]

Looks great! I think on the encoder side aside from mp3, wav we also support ogg, do we verify that it is one of these formats?

3:53 PM

[fjord-oai]

Looks great! I think on the encoder side aside from mp3, wav we also support ogg, do we verify that it is one of these formats?

chatted offline, we'll keep it open for now while we're figuring out exactly what is supported

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 519d780064

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".