AI multimodal input composer

mandatory

Unified composer for text + image + file + voice input. Drag-and-drop + paste hook + attach button + mic button. Upload preview chips with MIME validation gated by model capabilities. Voice mode transition to voice-mode.kmd (#121). Required for any AI chat surface.

Spec — AI multimodal input composer

Voice integration via voice-mode.kmd (#121) e voice/wake-word.kmd. Capability gating via model-selector.kmd (#113). Impl ticket voice-side: services/ai/ai#115 (já aberto).

Princípios

*ne composer, four modes*— text, image, file, voice. No produto separated UI per mode.
*rag~~and~~drop + paste + attach*— todo o composer aceita drops; paste from clipboard auto-attaches.
*IME validation gated by model*— accept only what active model can consume.
*oice as transition*— mic button enters dedicated voice mode (#121), não inline transcription.
*torage scoped*— uploads em path-prefix per workspace.

R1 — Anatomia

┌─────────────────────────────────────────────────┐
│ [📷 photo.jpg · 240KB · ✗]                      │  ← upload preview chips
│ [📄 spec.pdf · 1.2MB · ✗]                       │
├─────────────────────────────────────────────────┤
│ [🖼] [📎] [text input area...........] [🎙] [➤]│  ← composer row
└─────────────────────────────────────────────────┘

Slots:

Slot	Function
Preview chips	One per attached file; remove (✗) per chip
Image button 🖼	Open image picker (camera + gallery)
Attach button 📎	Open file picker (per model MIME whitelist)
Text input	Multi~~line; auto~~grow up to N lines; max-chars per model
Mic button 🎙	Enter voice mode (cross-link #121)
Send ➤	Submit; disabled if empty + no attachments

Drop zone: ENTIRE composer area accepts drops (visual highlight on dragover).

R2 — Upload preview chips

Per file attached:

┌────────────────────────────────────┐
│ [thumb] filename · MIME · size  ✗ │
└────────────────────────────────────┘

Thumb:

Image: actual thumbnail (96×96 max).
PDF: first page render (futuro; v1 = generic icon).
Audio: waveform mini.
Other: MIME icon.

Validation states:

State	Visual
Valid (model accepts)	Default chip
Invalid (model rejects MIME)	Red border + tooltip "Model X doesn't support {MIME}"
Uploading	Progress bar overlay
Failed	Error icon + Retry button

R3 — MIME validation per model

Current model capability (from model-selector.kmd #113):

Model capability	MIME accepted
Vision	imagepng, imagejpeg, imagewebp, imagegif
Documents	applicationpdf, textmarkdown, text/plain
Audio	audiomp3, audiowav, audioogg, audiom4a
Video	videomp4, videowebm
Files (generic text)	any text