feat: Add save parameter to image_search for downloading images to workspace

[devin-ai-integration]

Summary

Adds an optional save: Bool parameter to image_search (default false). When true, after fetching SerpAPI results and rendering the gallery as usual, the skill downloads each image's full-size original URL and persists it to the iCloud Drive workspace under downloads/images/<sanitized_query>_<index>.jpg.

Key design points:

• Additive, not a replacement — save: false (default) is identical to the old behavior. Gallery rendering is always intact.
• Concurrent downloads via DispatchGroup — each image downloads in parallel with a 15s timeout. Failures (network errors, non-2xx, invalid URLs) are logged and skipped; the search never crashes.
• Workspace integration — uses Workspace.shared.resolve + coordinatedWrite + coordinatedCreateDirectory so files land correctly on both iCloud and local-fallback backends.
• Result includes saved paths — the tool result appends a Saved N image(s) to workspace: section with workspace-relative paths (e.g. downloads/images/golden_retriever_1.jpg) so the model can reference them or call share_file.
• Filename sanitization — query → alphanumeric + _/-, capped at 60 chars. Extension derived from the source URL, defaulting to .jpg.
• System prompt updated — instructs the model to pass save: true when users ask to save/download/keep images.

Link to Devin session: https://app.devin.ai/sessions/254ea52d907048e19130d26ef462e47e

[devin-ai-integration]

Original prompt from API User

Enhance the image_search tool in the LoopHarness iOS app so that when Google Image Search results (via SerpAPI) are returned and rendered as a thumbnail gallery, the user can also SAVE those images to the iCloud Drive workspace.

Current behavior: image_search calls SerpAPI's Google Images endpoint, gets back results (title, url, thumbnail/originalUrl), and renders them as an inline tappable gallery in the chat. The images are not persisted anywhere.

Desired behavior:

1. Add an optional save boolean parameter to the image_search tool (default false). When save: true, after fetching results, download each image (the original/full-size URL if available, otherwise the thumbnail) and save it to the workspace under a folder like downloads/images/ with a sensible filename (e.g. image_search_<query>_<index>.jpg or based on the source domain).
2. When images are saved, include the saved file paths in the tool result so the model can reference them (and use share_file if needed).
3. If save is false (default), behavior is unchanged — just render the gallery as before.
4. Make sure to handle download failures gracefully (skip failed images, don't crash the whole search).
5. Keep the existing gallery rendering intact — saving is additive, not a replacement.

Look for the existing image_search implementation in the codebase (likely in a skills/tools directory). The SerpAPI integration should already be there. Build on top of it.

Add ashbhat as a reviewer on the PR.

Repository: theashbhat/LoopHarness

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
loop-harness	Ready	Preview, Comment	Jun 21, 2026 2:07pm