How to batch-watermark a folder of documents with an AI agent Leave feedback

Distribution sets arrive as folders, and marking them one by one is the kind of work that gets skipped under deadline. One prompt covers it.

Note

The commands and config snippets on this page are for the .NET build of the server — the only platform available today. Installation and client setup: MCP server for .NET. Other platforms will expose the same tools with their own launch command; everything else on this page applies unchanged.

Point GROUPDOCS_MCP_STORAGE_PATH at the folder (configuration). Files are resolved by name, so the agent passes proposal-01.pdf, not a path from your machine.

Add a diagonal CONFIDENTIAL watermark to every PDF in my documents folder, then list the files you produced.

The agent calls add_watermark per file. Each produces a new watermarked copy; the originals stay as they are.

Watermark each file with the client name and today’s date. Put the logo on the slide decks and a text mark on the PDFs. Mark everything except the ones that already have a watermark.

The last one is worth spelling out: the agent runs search_watermarks first and skips the files that already carry one — which is how you avoid a document with CONFIDENTIAL stamped twice at different angles.

• Ask for the list of produced files. A batch that reports only “done” leaves you guessing which copies to ship.
• Watch the failures. A mixed folder may contain a format the engine cannot watermark; the call fails with a message rather than silently skipping, so ask for failures to be listed explicitly.
• Mind metered usage. Under metered licensing each file is billed processing.
• Check the licence first. Evaluation-mode output carries evaluation limitations — a whole folder of them is a wasted run.

Approval flips the requirement. Same folder, opposite prompt:

Remove the DRAFT watermark from every file in this folder and keep the rest of the marks.

That is remove_watermarks with a textFilter — see Find and remove watermarks.