Render inline tool-generated images in chat#189
Conversation
- Extract tool-result image blocks in ACP and Claude canonical mappers - Add ImageView with lightbox, clipboard copy, and save-as IPC handlers - Extend tool-call payload schema and keep image tool calls discrete on reload - Reuse ImageLightbox for shared fullscreen preview in chat and composer
|
The latest updates on your projects. Learn more about Vercel for GitHub. 1 Skipped Deployment
|
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 48b5338fde
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".
| for (const value of record.images) { | ||
| if (typeof value !== "string" || value.length === 0) continue; | ||
| const classification = classifyCandidate(value); | ||
| if (classification) return { value, classification }; |
There was a problem hiding this comment.
When an ACP/Claude tool result contains more than one image block, the mappers now preserve all of them on payload.images, but this resolver returns as soon as it finds the first renderable entry. In that scenario the chat card, lightbox, copy/download actions, and persisted row only expose the first image while silently dropping the rest; consider resolving and rendering the full images array instead of a single candidate.
Useful? React with 👍 / 👎.
1 participant
imageGeneration, Claude/ACP image tools) inline in chat instead of folding them into generic tool-call rowstoolCallPayload.imagesso provider-native picture data reaches the rendererImageViewand shared image-source resolution for base64/data:payloads, with copy, download, and lightbox; render image blocks in assistant messages toosaveImageFile,copyImageToClipboard) to save or copy generated images from the UI