Skip to content

Latest commit

 

History

History

nutrient

Folders and files

README.md

Nutrient

Extract data from PDFs, Office documents, images and scans with your Nutrient account (Nutrient DWS, Nutrient's document API). Nutrient shows where on the page it found the values, and it warns you about values to check. You can also convert documents to Markdown and classify them. With files on your computer, as in Claude Code and Codex, you can also make searchable PDFs, convert files (for example, Word to PDF), redact them and add your organization's digital signature. The plugin bundles the Nutrient connector with a skill that tells Claude, ChatGPT or Codex when to use it, so a request such as "Extract the invoice number and total" runs through Nutrient even when the file is attached to the chat.

Use it

  1. Add the plugin, then connect the Nutrient connector and sign in with your Nutrient account. The tools appear once you're signed in.
    • In Claude Code, install it with /plugin install nutrient --marketplace PSPDFKit-labs/nutrient-mcp-plugins, then run /mcp, choose plugin:nutrient:nutrient and sign in.
    • In Codex, install it with codex plugin marketplace add PSPDFKit-labs/nutrient-mcp-plugins and codex plugin add nutrient@nutrient, sign in with codex mcp login nutrient, then start a new session.
  2. Attach a PDF, Office document or image, point to a file on disk or an HTTPS link, and ask to parse it, convert it to Markdown, extract fields or tables, or classify it.
  3. The assistant sends the file to your Nutrient account and returns the result: Markdown with layout, structured fields with citations for the values it finds, or labels with scores. Where the Nutrient panel appears, you can review each value beside its source on the page.

In ChatGPT, Nutrient fetches a file attached to the chat from the link ChatGPT provides. In Claude, for Claude to upload an attached file itself, mcp.nutrient.io must be under Settings, Capabilities, Additional allowed domains (set by an Owner on Team and Enterprise plans); otherwise Claude asks you to drop the file in the Nutrient panel.

Data

The plugin sends the documents you ask the assistant to process, and the instructions for each run, to your Nutrient DWS account through mcp.nutrient.io; it sends them nowhere else. To send a file from your computer, or one attached to the chat, the assistant runs a one-time curl command that the connector returns: it uploads only the files you named to mcp.nutrient.io and expires within five minutes. Uploaded files are kept for 24 hours. DWS keeps each run for your plan's retention period, and each run uses your DWS credits. Opening a document in the Nutrient panel's full-screen review stores a copy in DWS Viewer, which stays until you delete it and counts toward your plan's document limit. Summaries and questions about a document stay with the assistant's own reading unless you ask for one of these operations.

Privacy policy: https://www.nutrient.io/legal/privacy/

Troubleshooting

  • HTTP 403 "Host not in allowlist" when uploading: allow mcp.nutrient.io in the app's network settings (in Claude: Settings, Capabilities, Additional allowed domains), or drop the file in the Nutrient panel when it asks for it.
  • The connector asks you to connect or reconnect: sign in again from the plugin's or the connector's settings.
  • permission_denied: your Nutrient account doesn't include the product the request needs, such as Data Extraction or Processor. Check your plan in the DWS dashboard.
  • Not enough credits: a run stops before processing when your DWS credits run out; add credits in the DWS dashboard.
  • Review says DWS Viewer isn't set up for your organization: your Nutrient organization hasn't chosen where DWS Viewer stores documents yet. Someone in the organization can choose a region once in the DWS dashboard, under Viewer API, Documents; it can't be changed later. The panel's Open DWS dashboard button goes there.

Support