[pipecat-docs-bot]

Automated documentation update for pipecat PR #6594.

Changes

• api-reference/server/classifiers/overview.mdx: ask()/yes_no()/choice()/score() take inputs=; images= marked deprecated (1.16.0, removed 2.0.0); the audio= argument is replaced by inputs=; added Files section with ClassifierFile and file_types property; updated ClassifierError, import block, and "Writing a Classifier" (_ask() takes inputs).
• pipecat/learn/classifiers.mdx, classifiers/clef.mdx, llm.mdx, openai-decisions.mdx, jev.mdx, hume-audio.mdx: examples and prose use inputs=; Hume notes that a call takes exactly one clip.
• Review: deprecation note format in overview.mdx; hume-audio.mdx usage line says a call needs exactly one clip in inputs.

Gaps identified

None. No built-in classifier takes files (file_types is empty), so no provider page documents file support. PerceptionReader changes are internal.

Automated review

• overview.mdx: deprecation of images used a <Warning> block unlike other pages; changed to the "Deprecated since 1.16.0, removed in 2.0.0." format.
• hume-audio.mdx: "Every question needs a clip" contradicted the new rule (exactly one clip in inputs); reworded.
• Reference checker: no findings on the pages this update changes (only an unrelated jambonz default mismatch, left alone).
• Claims, coverage, other pages checked against the code: no other problems found.

[github-actions]

🔍 Mintlify preview for this branch: https://daily-docs-pr-6594.mintlify.site