Details:

Homeworks currently exposes visit media through the MCP/API, including the media ID, filename, content type, thumbnail URL, and file URL. However, connected AI agents such as ChatGPT cannot directly retrieve the underlying image or video file for visual analysis.

Use case: our field technicians upload photos and videos to customer visits. I want the connected AI agent to be able to review those files directly from the visit, analyze job-site conditions, compare before/after photos, review completed work, identify potential damage or quality issues, and analyze videos without having to manually download each file from Homeworks and upload it into ChatGPT.

Requested capability: expose visit media as a downloadable MCP resource/file, authenticated binary download action, or another mechanism that allows the connected AI client to ingest the actual photo/video file rather than only receiving its URL and metadata.

The existing event.media and beforeAndAfters media data already exposes the records, so the missing piece appears to be authenticated file retrieval through MCP.

This would make field photos and videos significantly more useful for QA, estimating, damage review, job documentation, and AI-assisted field operations.