Here are some features that I can think of that are useful
1. Image tag and metadata "hash"
Image data hash, so if you were to download metadata from the same file twice, and the metadata has not changed, the hash would be the same, telling the user to not download it again and waste precious bandwidth.
2. IPFS H2H API
IPFS operational API, so that you can create folder or file hash of any search results, and be able to pin or unpin them at will. This will be useful in the long term when people use it for Hydrus-2-Hydrus (H2H) P2P file share.
3. De-duplication API
internal APIs for file de-duplication software like Pippy360 for images and Chromaprint for music, and then allow external APIs be called for IQDB-like file search and hash download capabilities. (requires APIs for uploads first)
4. Headless Drone API
Headless "master+drone" Hydrus APIs, that allows one "master" to distribute downloader scripts and "download orders" to them, and the "drones" will return either downloaded images and tags, or an error message, back to the "master"