Results come back a window at a time, with the count and start index on the call and safe search alongside them, so a broad query can be worked through in batches. Saving takes the chosen result's URL and a path relative to the workspace and creates the target folder when it is not there. There is no editing, resizing or metadata step here — the image lands as it was found.
A two-tool image server: a paged online image search with a safe-search setting, and a downloader that stores a result into the current workspace.
- Paged image search — a query with a result count and a start index, and a safe-search setting on the same call.
- Storage of an image from its URL to a folder relative to the current workspace, with the target path created automatically when it does not exist.
An API key and a search engine ID for image search, supplied as API_KEY and SEARCH_ENGINE_ID, plus a workspace path the server may write into.
One command plus a key — npx -y @srigi/mcp-google-images-search, then supply credentials