[FEATURE]: Support drag-and-drop / paste of images in the question tool UI #31791

Description

When the agent invokes the question tool (MCQ / structured question UI), there is currently no way to attach images or screenshots while answering. Drag-and-drop and clipboard paste work (or are planned) for the main chat prompt, but the question interface appears to be text/selection-only.

This is a common workflow gap: the agent asks a clarifying question, and the user wants to respond with a screenshot (e.g. "which layout do you mean?" → drop a UI screenshot) without dismissing the question modal or switching back to the main prompt.

Expected behavior

While the question tool UI is open, users should be able to:

• Drag and drop image files (PNG, JPG, GIF, WebP) into the question interface
• Paste screenshots from the clipboard (Cmd/Ctrl+V)
• See attached images as thumbnails/previews before submitting answers
• Have attached images included in the question response sent back to the agent

Actual behavior

• Dropping images onto the question UI does nothing (no dropzone, no attachment)
• Pasting a screenshot while focused on the question UI does not attach it
• Users must dismiss or work around the question flow and attach images via the main prompt instead

Motivation / use cases

1. Agent asks "Which error screen are you seeing?" → user drops a screenshot inline
2. Agent asks "Which of these UI variants should we match?" → user attaches reference images
3. Debugging sessions where visual context is faster than describing UI state in text

Related issues

These cover image attach for the main prompt, but not the question tool UI specifically:

• Feature request: Paste to attach image #906 — Feature request: Paste to attach image
• [FEATURE] Ctrl+V paste images from clipboard into chat (Windows desktop app) #26111 — Windows desktop Ctrl+V + drag-drop
• [Bug] Images silently discarded with local Ollama vision models + TUI lacks drag-and-drop #26780 — TUI drag-and-drop support
• [FEATURE]: Add attach button in terminal TUI for images/screenshots #20169 — TUI attach button when drag-drop fails
• [FEATURE]: Desktop should accept more dropped file types in prompt input #18143 — Desktop dropped files in prompt

Question tool UI issues (layout/behavior, not image attach):

• [Bug]: Question Tool popup doesn't appear casually, and opencode auto-selects an option then continue outputting or editing #31383 — Question Tool popup doesn't appear casually
• Long question tool makes opencode unusable #11367 — Long question tool makes opencode unusable
• [FEATURE]: Generic UI Intent Channel for cross-client plugin-driven UX #6330 — Generic UI Intent Channel (broader cross-client question UI architecture)

Suggested implementation notes

• Reuse existing prompt attachment / dropzone infrastructure where possible (prompt.dropzone.label i18n strings already exist for desktop/web)
• Extend the question tool response schema to support optional image attachments per answer or as freeform context
• Ensure parity across Desktop, Web, and TUI clients (or document platform limitations)

OpenCode version

Latest (observed June 2026)

Steps to reproduce

1. Start a session where the agent invokes the question tool
2. When the question UI appears, try dragging a PNG/JPG screenshot onto it
3. Alternatively, copy a screenshot to clipboard and press Cmd/Ctrl+V while the question UI is focused
4. Observe that no image is attached and the answer can only be text/selection-based

Operating System

macOS (likely affects Desktop/Web/TUI generally)

Terminal

N/A (Desktop app / question modal)