Connect the Google Docs MCP server to your AI agent Open source, MIT, on npm. Setup instructions for Claude, Cursor and Codex are in the repository; adoption questions — Telegram. GitHub repository Telegram

Google Docs: Read document as text — MCP tool

Google Docs MCP tool: Returns the document as compact readable blocks with edit-ready indexes.

Technical name: read_document_text

What task it solves

I want to read a document's content as plain text.

Returns the document as compact readable blocks with edit-ready indexes.

When to use it

The default reading tool: use it before summarizing, quoting or editing a document. Every paragraph comes with its start/end indexes — exactly the coordinates that insert_text, replace_range, delete_range and the style tools take — plus heading styles, bullet flags and table cell text.

What to provide

  • document_idrequired.
  • tab_id — optional. Limit the output to one tab (ids from list_tabs).

What it returns

Per tab: an array of blocks — paragraphs (text, start, end, heading style, bullet) and tables (rows, columns, cells[row][column] plain text). Inline images appear as [image:<objectId>] placeholders.

What changes in Google Docs

Nothing — the tool reads Google Docs data and does not change it.

Example request

Read the launch plan document and summarize the open questions section.

Errors and limitations

Indexes go stale after ANY edit — re-read between mutations. Footnotes, headers and footers are not part of the body blocks; fetch them via get_document.

Access also depends on token permissions, quotas, and upstream API limits.

Related MCP tools

Technical details