Atlas / MCP servers / securityronin / Docx

DocxSAFE

Name: Docx
Author: securityronin

mcp/securityronin/docx-2

MCP server by securityronin · ai tools · comments · document editing · Audited 2026-10-08

MCP server for reading and editing Word (.docx) documents with track changes, comments, footnotes, and structural validation

Verdict

SAFE

Grade

Trust score

85 /100

Exposed tools

200 72r · 119w · 27d

Transport

—

License

MIT

Stars

Overview

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

[](https://pypi.org/project/docx-mcp-server/) [](https://pypi.org/project/docx-mcp-server/) [](https://opensource.org/licenses/MIT) [](https://github.com/SecurityRonin/docx-mcp/actions/workflows/ci.yml) [](https://github.com/SecurityRonin/docx-mcp) [](https://github.com/sponsors/h4x0r) [](https://safeskill.dev/scan/securityronin-docx-mcp)

Give your AI coding agent the ability to create, read, and edit Word documents. Edits appear as tracked changes in Microsoft Word — red strikethrough for deletions, green underline for insertions — and after any revision session your agent can produce an email-ready change log listing every insertion, deletion, and replacement. Or compare two separate files to get the same output automatically.

Who This Is For

Professionals who produce Word deliverables and want their AI agent to handle the document work directly:

Legal — contract review with tracked redlines, batch clause replacement across templates, comment annotations explaining each change, footnote management
Security & Penetration Testing — generate pentest reports from markdown findings, merge appendices from multiple engagements, add executive-summary comments, remove DRAFT watermarks before delive

Read from source at commit 3ab2b904890bOBSERVED · 2026-10-08

Connect

Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.

claude-code

claude mcp add docx-mcp-server -- uvx docx-mcp-server

claude-desktop

{
  "mcpServers": {
    "docx-mcp-server": {
      "command": "uvx",
      "args": [
        "docx-mcp-server"
      ]
    }
  }
}

Exposed tools (200)

72 read · 119 write · 27 destructive. Blast radius: 27 tools can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.

Tool	Risk	Description
`accept_all_changes`	write	Accept all tracked changes in document order.
`accept_change`	read	Accept a single tracked change by its change_id.
`accept_changes`	destructive	Accept tracked changes — keep insertions, remove deletions. Empty author = all.
`add_bookmark`	write	Add a named bookmark wrapping the specified paragraph.
`add_column_to_table`	write	Add a new column to every row of a table. First row gets header_text.
`add_comment`	write	Add a comment anchored to a paragraph.
`add_content_control`	write	Wrap a paragraph in an SDT content control.
`add_cross_reference`	write	Add a cross-reference link from one paragraph to another.
`add_endnote`	write	Add an endnote to a paragraph.
`add_equation`	write	Insert a LaTeX equation as OMML. Requires: pip install latex2mathml.
`add_field`	write	Insert a Word field at end of paragraph.
`add_footnote`	write	Add a footnote to a paragraph. url, if provided, is rendered as a hotlink.
`add_footnote_ref`	write	Add a subsequent reference to an existing footnote without creating a new definition.
`add_hyperlink`	write	Append an external hyperlink at the end of a paragraph.
`add_internal_link`	write	Append an internal anchor hyperlink (w:anchor) at the end of a paragraph.
`add_list`	write	Apply list formatting to paragraphs (bullet or numbered).
`add_page_break`	write	Insert a page break after a paragraph.
`add_section_break`	write	Add a section break at a paragraph. break_type: nextPage/continuous/evenPage/oddPage.
`add_table`	write	Insert a new table after a paragraph with tracked insertion.
`add_table_row`	write	Add a row to a table with tracked insertion. row_idx=-1 appends.
`apply_style_to_range`	write	Apply a style to a list of paragraphs by their paraIds.
`audit_document`	write	Run a comprehensive structural audit of the document.
`check_accessibility`	read	Scan the document for accessibility issues.
`clear_run_formatting`	destructive	Remove all character formatting from a run, causing it to inherit paragraph/style defaults.
`close_document`	read	Close a document and clean up temporary files.
`compare_contracts`	read	Clause-aware diff between the open contract and another .docx file.
`compare_documents`	read	Diff two DOCX files and produce a tracked-change document.
`convert_to_pdf`	read	Convert the open document to PDF using LibreOffice headless.
`copy_document`	write	Save a complete snapshot of the open document to a new path.
`copy_style`	read	Deep-copy an existing style under a new name.
`copy_table`	write	Deep-copy a table and insert the copy immediately after the original.
`create_document`	write	Create a new blank .docx document (or from a .dotx template).
`create_from_markdown`	write	Create a new .docx document from markdown content.
`create_multilevel_list`	write	Create a multilevel list in numbering.xml. Each level dict: {num_fmt, lvl_text, indent, hanging, style?}.
`create_style`	write	Create a new style in the document.
`csv_to_table`	write	Insert a table from CSV text.
`delete_column_from_table`	destructive	Delete a column (0-based) from every row of a table.
`delete_comment`	destructive	Delete a comment and remove its range markers from the document.
`delete_content_control`	destructive	Remove an SDT content control wrapper, keeping its content in place.
`delete_custom_property`	destructive	Delete a custom document property by name.
`delete_endnote`	destructive	Delete an endnote and its in-body reference.
`delete_field`	destructive	Remove a complete complex field (begin through end runs) from the document.
`delete_footer`	destructive	Delete a footer by location: default, first, or even.
`delete_footnote`	destructive	Delete a footnote and its in-body reference.
`delete_header`	destructive	Delete a header by location: default, first, or even.
`delete_image`	destructive	Remove the drawing containing the image with the given rId from the document.
`delete_paragraph`	destructive	Delete the paragraph with the given paraId.
`delete_section_break`	destructive	Remove a section break from a paragraph.
`delete_style`	destructive	Delete a style from the document.
`delete_table`	destructive	Delete a table by index (0-based). Raises IndexError if out of range.
`delete_table_row`	destructive	Delete a table row with tracked changes.
`delete_text`	destructive	Delete text from a paragraph.
`demote_list_item`	read	Increase the list indentation level (ilvl) of a paragraph by 1, maximum 8.
`diff_to_text`	read	Compare two separate DOCX files and produce a tracked-change DOCX plus a plain-text summary.
`duplicate_table_row`	write	Deep-copy a table row and insert the copy immediately after it.
`edit_header_footer`	write	Edit text in a header or footer.
`export_markdown`	read	Export the open document as Markdown.
`export_session_script`	write	Write session operations as a Python replay script.
`fill_template`	read	Fill SDT content controls from data dict. Keys match w:tag values.
`find_replace_formatted`	read	Find all occurrences of a string and replace with formatted text.
`flatten_document`	destructive	Accept all tracked changes and remove all revision markup.
`generate_change_summary`	read	Summarise tracked changes already present in the open document as an email-ready .txt.
`generate_list_of_figures`	write	Insert a List of Figures field (requires SEQ Figure captions).
`generate_list_of_tables`	write	Insert a List of Tables field (requires SEQ Table captions).
`generate_privilege_log`	read	Generate a privilege log DOCX from document metadata.
`generate_redaction_log`	write	Write a DOCX table of all redactions made this session.
`generate_toc`	read	Generate a Table of Contents from document headings.
`generate_tof`	write	Insert a Table of Figures field block after the paragraph with para_id.
`generate_tot`	write	Insert a Table of Tables field block after the paragraph with para_id.
`get_alt_text`	read	Get the alt text and title for an image by 0-based index.
`get_body_text`	read	Return the full accepted-view text of the document.
`get_bookmarked_text`	read	Get the text content within a named bookmark.
`get_cell_text`	read	Return text content of a specific cell.
`get_comments`	read	List all comments with their ID, author, date, and text.
`get_content_control`	read	Return details of a single content control by its w:id.
`get_content_controls`	read	List all SDT content controls in the document.
`get_custom_properties`	read	Get custom document properties from docProps/custom.xml.
`get_document_info`	read	Get overview stats: paragraph count, headings, footnotes, comments, images.
`get_document_outline`	read	Return a flat list of headings as a document outline.
`get_endnotes`	read	Get all endnotes with their ID and text content.
`get_equations`	read	Return all equations in the document as OMML XML strings.
`get_field`	read	Return details of a single field by field_id.
`get_footnotes`	read	List all footnotes with their ID and text content.
`get_headers_footers`	read	Get all headers and footers with their text content.
`get_headings`	read	Get the document heading structure with levels, text, and paraIds.
`get_images`	read	Get all embedded images with rId, filename, content type, and dimensions.
`get_lists`	read	Return all list definitions from numbering.xml.
`get_paragraph`	read	Get the full text and style of a specific paragraph by its paraId.
`get_paragraph_format`	read	Read all formatting attributes of a paragraph.
`get_properties`	read	Get core document properties (title, creator, subject, dates, revision).
`get_reading_time`	read	Estimate reading time for the open document.
`get_runs`	read	Get all runs in a paragraph with their formatting properties.
`get_sections`	read	List all sections in the document with their properties.
`get_session_log`	read	Return all operations performed this session as replayable JSON.
`get_statistics`	read	Return document statistics for the open document.
`get_style`	read	Get details of a single style by name or styleId (case-insensitive).
`get_styles`	read	Get all defined styles with ID, name, type, and base style.
`get_table`	read	Get structured info for a single table by zero-based index.
`get_tables`	read	Get all tables with row/column counts and cell text content.
`get_theme_colors`	read	Return the named color slots from word/theme/theme1.xml.
`get_tracked_changes`	read	Return all pending tracked changes (insertions and deletions) as a JSON list.
`get_word_count`	read	Return the word count of the open document body.
`insert_bar_chart`	write	Insert a native bar chart (no Excel required).
`insert_blockquote`	write	Insert a blockquote paragraph after the given paragraph.
`insert_caption`	write	Insert a caption paragraph after the specified paragraph.
`insert_code_block`	write	Insert a code-block paragraph after the given paragraph.
`insert_date_field`	write	Insert a DATE field at the end of a paragraph.
`insert_floating_image`	write	Insert a floating (anchored) image. wrap: square\|topbottom\|none.
`insert_if_field`	write	Insert a Word IF conditional field at the end of a paragraph.
`insert_image`	write	Insert an image into the document after a paragraph.
`insert_line_chart`	write	Insert a native line chart.
`insert_merge_field`	write	Insert a MERGEFIELD (mail merge) field at the end of a paragraph.
`insert_page_number_field`	write	Insert a PAGE field at the end of a paragraph.
`insert_paragraph`	write	Insert a new paragraph after the paragraph with the given paraId.
`insert_pie_chart`	write	Insert a native pie chart (single series, fixed 14x9 cm).
`insert_sequence_field`	write	Insert a SEQ (sequence) field for figure/table numbering.
`insert_text`	write	Insert text into a paragraph.
`insert_text_box`	write	Insert an inline text box after the paragraph with para_id.
`insert_watermark`	write	Insert a VML watermark into the document
`list_bookmarks`	read	List all bookmarks in the document.
`list_comment_threads`	read	List all comment threads (root comments with their replies).
`list_fields`	read	List all fields in the document with their codes and cached values.
`list_hyperlinks`	read	List all hyperlinks in the document.
`list_parts`	read	List all XML parts (files) in the open DOCX zip.
`list_template_fields`	read	List all SDT template fields (tag, label, type) in the document.
`lock_content_control`	read	Lock a content control to prevent editing.
`merge_cells`	write	Merge a rectangular range of cells. Horizontal: gridSpan. Vertical: vMerge.
`merge_documents`	write	Merge another DOCX document
`merge_review_rounds`	write	Merge tracked changes from N reviewer copies into the open document.
`modify_cell`	write	Modify a table cell.
`open_document`	read	Open a .docx file for reading and editing.
`promote_list_item`	read	Decrease the list indentation level (ilvl) of a paragraph by 1, minimum 0.
`read_part`	read	Read raw XML of any DOCX part (e.g.
`redact_text`	destructive	True redaction: remove text and replace with black rectangle. Use exact_text or pattern.
`reject_all_changes`	write	Reject all tracked changes in document order.
`reject_change`	read	Reject a single tracked change by its change_id.
`reject_changes`	destructive	Reject tracked changes — remove insertions, restore deleted text.
`remove_bookmark`	destructive	Remove a bookmark by name (keeps paragraph content).
`remove_hyperlink`	destructive	Remove a hyperlink wrapper, preserving the text runs inside.
`remove_watermark`	destructive	Remove VML watermarks (e.g., DRAFT) from all document headers.
`replace_text`	read	Replace text in a paragraph.
`reply_to_comment`	read	Reply to an existing comment (creates a threaded reply).
`resolve_comment`	read	Mark a comment as resolved (sets w15:done=
`restart_numbering`	write	Restart list numbering at a paragraph. Adds lvlOverride with startOverride.
`sanitize_metadata`	write	Write a sanitized copy of the open document to output_path.
`save_document`	write	Save all changes back to a .docx file.
`scrub_pii`	read	[EXPERIMENTAL] Detect and redact PII from the open document using Presidio + spaCy NER.
`search_text`	read	Search for text across the document body, footnotes, and comments.
`set_alt_text`	write	Set the alt text (and optionally title) on an image by 0-based index.
`set_cell_shading`	write	Set background shading fill color on a table cell.
`set_cell_vertical_alignment`	write	Set vertical alignment of a table cell: top, center, or bottom.
`set_cell_width`	write	Set the width of a table cell in millimetres (stored as DXA).
`set_character_position`	write	Set vertical character position (raised/lowered) for a specific run.
`set_character_spacing`	write	Set character spacing (tracking) for a specific run in a paragraph.
`set_column_widths`	write	Set column widths in cm. len(widths_cm) must match column count.
`set_content_control_value`	write	Update the value/text of a content control by its tag.
`set_custom_property`	write	Set (upsert) a custom document property.
`set_different_first_page`	write	Enable or disable a different first-page header/footer for a section.
`set_document_language`	write	Set the default document language.
`set_document_protection`	write	Set document protection. edit: trackedChanges/comments/readOnly/forms/none.
`set_formatting`	write	Apply character formatting to text with tracked-change markup.
`set_header_row`	write	Mark the first row as a repeating header row.
`set_image_alt_text`	write	Set accessibility alt text and title on an embedded image.
`set_image_border`	destructive	Set or remove a border on an embedded image.
`set_image_size`	write	Resize an embedded image by updating its EMU extent attributes.
`set_keep_lines_together`	write	Keep all lines of this paragraph on the same page.
`set_keep_with_next`	write	Keep this paragraph on the same page as the next paragraph.
`set_line_spacing`	write	Set line spacing and paragraph spacing.
`set_odd_even_headers`	write	Enable or disable different odd/even page headers globally.
`set_page_break_before`	destructive	Force a page break before this paragraph.
`set_page_margins`	write	Set page margins from millimetre values.
`set_page_orientation`	write	Set page orientation, swapping width/height dimensions if needed.
`set_page_size`	write	Set page size from millimetre values.
`set_paragraph_border`	write	Set borders on one or more sides of a paragraph.
`set_paragraph_indentation`	write	Set indentation on a paragraph.
`set_paragraph_shading`	write	Set background shading on a paragraph.
`set_properties`	write	Set core document properties. Empty string = unchanged.
`set_row_height`	write	Set row height in millimetres. rule: exact, atLeast, or auto.
`set_run_color`	write	Set the font color of a specific run in a paragraph.
`set_run_font`	write	Set the font of a specific run (zero-based index) in a paragraph.
`set_run_highlight`	write	Set highlight color of a specific run in a paragraph.
`set_run_language`	write	Set the language on a run for spell-checking purposes.
`set_run_size`	write	Set the font size of a specific run in a paragraph.
`set_run_strikethrough`	write	Set strikethrough on a specific run in a paragraph.
`set_run_subscript`	write	Set subscript vertical alignment on a specific run in a paragraph.
`set_run_superscript`	write	Set superscript vertical alignment on a specific run in a paragraph.
`set_run_underline`	write	Set underline style on a specific run in a paragraph.
`set_section_columns`	write	Set the number of columns in a section.
`set_section_properties`	write	Modify section properties (page size, orientation, margins). 0/empty = unchanged.
`set_table_alignment`	write	Set table alignment: left, center, or right.
`set_table_borders`	write	Set borders on all six sides of a table (top, bottom, left, right, insideH, insideV).
`set_table_style`	write	Apply a named table style (e.g. TableGrid, LightShading-Accent1) to a table.
`set_text_case`	write	Set text case transformation on a run.
`set_theme_color`	write	Update a named color slot in the document theme.
`set_track_changes`	write	Enable or disable revision tracking in the document.
`set_widow_control`	write	Enable widow/orphan control for this paragraph.
`sort_table`	read	Sort the non-header rows of a table by the text content of a column.
`split_document`	read	Split the open document into multiple DOCX files, one per heading section.
`split_table`	read	Split a table into two tables at the given row index.
`suppress_numbering`	destructive	Remove list numbering from a paragraph by setting numId to 0.

Trust audit

SAFEgrade B · trust 85/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

Layer	What it checks	Result
L0	Provenance & inventory	PASS
L1	Static analysis of the code	PASS
L2	Instruction surface (what it tells the agent)	PASS
L3	Class-specific surface	WARN
L4	Behavioural (sandbox)	SKIPPED

What the source does

Filesystem: declared (6 observation(s))
Network: none-observed
Shell: none-observed
Dependencies: pinned
Secrets in source: none-found

Findings (23)

MEDIUMFilesystem / path · mcp.destructive_tools · CWE-22, CWE-59

accept_changes, clear_run_formatting, delete_column_from_table, delete_comment, delete_content_control, delete_custom_property, delete_endnote, delete_field, delete_footer, delete_footnote, delete_hea

Why it matters. 27 tool(s) can delete or overwrite

Fix. prefer a read-only mode or scoped tokens; the page states the blast radius

LOWInventory / provenance · inv.binary · CWE-1104

tests/fixtures/lo_chart.docx

lo_chart.docx