Google Docs¶

Create, read, and programmatically build documents with paragraphs, headings, lists, tables, images, and formatting.

Tools¶

Tool Name	Description
`docs_create_document`	Create a new document with a given name.
`docs_get_document`	Get document metadata plus paragraph text, bounded by proxy paragraph and character limits.
`docs_insert_paragraph`	Insert a paragraph with optional text and heading level (NORMAL, HEADING1-HEADING6).
`docs_update_paragraph`	Update an existing paragraph's text and/or heading level by index.
`docs_delete_paragraph`	Delete a paragraph by index.
`docs_set_text`	Replace the entire document body with new text.
`docs_replace_text`	Find and replace text across the entire document.
`docs_insert_list`	Insert a bulleted or numbered list from an array of items.
`docs_insert_table`	Insert a table from a 2D array of values (first row auto-bolded as header).
`docs_insert_image`	Insert an image from a public URL.
`docs_insert_page_break`	Insert a page break divider.
`docs_insert_horizontal_rule`	Insert a horizontal rule divider.
`docs_format_text`	Format text by search pattern — bold, italic, underline, strikethrough, font, size, colors, and links.
`docs_set_header`	Set the document header text (empty string to clear).
`docs_set_footer`	Set the document footer text (empty string to clear).
`docs_get_as_json`	Get the document as structured JSON with formatting and structure, bounded by proxy response-size limits.
`docs_get_page_setup`	Read page size and margins in points.
`docs_update_page_setup`	Update page size and margins in points.
`docs_list_bookmarks`	List bookmarks with IDs and paragraph positions when available.
`docs_create_bookmark`	Create a bookmark at a paragraph character offset.
`docs_delete_bookmark`	Delete a bookmark by ID.
`docs_list_named_ranges`	List named ranges, optionally filtered by name.
`docs_create_named_range`	Create a named range around a paragraph or partial paragraph span.
`docs_delete_named_range`	Delete a named range by ID.
`docs_list_table_of_contents`	List existing table-of-contents elements with child indexes and text previews.
`docs_batch`	Execute multiple operations (up to 20) in a single round-trip.

Key Features¶

Heading levels — insert_paragraph and update_paragraph support six heading levels (HEADING1 through HEADING6) plus NORMAL body text. Build structured, navigable documents with a proper outline.
Paragraph-based API — The document model is paragraph-centric. Each paragraph has an index (0-based) and can be independently inserted, updated, or deleted.
Pattern-based formatting — format_text finds specific text patterns anywhere in the document and applies formatting (bold, italic, colors, font, links) only to matches.
Rich element insertion — Insert images (from URLs), tables (with auto-bolded headers), bulleted or numbered lists, page breaks, and horizontal rules — all with append/prepend positioning control.
Structural controls — Page setup, bookmarks, named ranges, and existing table-of-contents elements are exposed as constrained tools rather than raw Docs API requests.
Batch operations — Use docs_batch to chain up to 20 document operations in a single round-trip.

Examples¶

Build a structured report from scratch:

# Create the document
docs_create_document({ name: "Monthly Status Report" })

# Add a title
docs_insert_paragraph({
  documentId: "<id>",
  heading: "HEADING1",
  text: "Monthly Status Report — June 2026"
})

# Add a key points list
docs_insert_list({
  documentId: "<id>",
  items: ["Deployed v2.3 to production", "Resolved 12 customer tickets", "Started Q3 roadmap planning"],
  listType: "BULLET"
})

# Add a section with table
docs_insert_paragraph({
  documentId: "<id>",
  heading: "HEADING2",
  text: "Team Metrics"
})
docs_insert_table({
  documentId: "<id>",
  values: [["Team", "Velocity", "Bugs"], ["Frontend", "42", "3"], ["Backend", "38", "7"]]
})
// First row auto-bolded as header

Add document structure and navigation:

docs_update_page_setup({
  documentId: "<id>",
  pageWidth: 612,
  pageHeight: 792,
  marginTop: 72,
  marginBottom: 72,
  marginLeft: 72,
  marginRight: 72
})

docs_create_bookmark({
  documentId: "<id>",
  paragraphIndex: 3,
  offset: 0
})

docs_create_named_range({
  documentId: "<id>",
  name: "executive-summary",
  paragraphIndex: 3
})

docs_list_table_of_contents({ documentId: "<id>" })

Format specific terms throughout a document:

docs_format_text({
  documentId: "<id>",
  findText: "v2.3",
  bold: true,
  foregroundColor: "#0066CC"
})
// Every occurrence of "v2.3" is now bold and blue

Update a specific section without rewriting the whole document:

# Get document to find paragraph indices
docs_get_document({ documentId: "<id>" })

# Update paragraph 3 to change its text and heading level
docs_update_paragraph({
  documentId: "<id>",
  paragraphIndex: 3,
  text: "Q3 Roadmap Kickoff — Completed",
  heading: "HEADING2"
})

Limits & Considerations¶

The paragraph index model requires knowing the current structure. After inserting or deleting paragraphs, all subsequent indices shift. Use docs_get_document to inspect the current layout before indexed operations.
Images must be publicly accessible URLs; local file uploads are not supported directly.
docs_set_text replaces the entire document body — use with caution on existing documents. Prefer insert_paragraph and replace_text for targeted changes.
docs_get_document is limited to 500 paragraphs and 500,000 characters; docs_get_as_json is limited to a 1,000,000-character JSON payload.
docs_delete_bookmark and docs_delete_named_range require confirm: true because they remove document metadata.
Bookmarks and named ranges are anchored to paragraph indices. Use docs_get_document, docs_list_bookmarks, or docs_list_named_ranges after structural edits because indices can shift.
Google Docs table-of-contents elements can be inspected when already present, but Apps Script and the Docs API do not expose a supported creation method through this service surface.
format_text applies formatting to all occurrences of the search text. For single-instance formatting, ensure the search text is unique within the document.
Headers and footers accept plain text only; rich formatting in headers/footers is not supported through this tool surface.