Any supported viewer URL
NLS, NLW, handle.net links and most IIIF-backed viewers — plus a generic fallback that sniffs unknown sites for a manifest.
Digital libraries publish manuscripts and rare books through IIIF — but AI assistants usually can’t get at them: sandboxed clients can’t fetch arbitrary image URLs, and finding the IIIF manifest behind a viewer page takes repository-specific knowledge. iiiftools is a managed MCP server that closes both gaps. Give it any supported viewer URL and it resolves the manifest, downloads the page images to temporary storage, and hands them to your assistant — inline, or as short-lived links.
Any supported viewer URL
NLS, NLW, handle.net links and most IIIF-backed viewers — plus a generic fallback that sniffs unknown sites for a manifest.
Rules, not guesswork
Per-repository resolution rules live in a database and are editable in the dashboard — a shared source of truth linking repositories to their IIIF endpoints.
Temporary by design
Images are held only for the working session: released on request, expired on a TTL. Nothing is archived server-side.
No vision server-side
iiiftools does no transcription or image analysis — it is the plumbing that gets the page in front of the model doing the reading.