AI agents use add_paragraph to create or update resources in Docx — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Docx environment.
Adding a paragraph to a document is a reversible write operation that creates or modifies data without destroying information. The blast radius is medium because a compromised agent could inject malicious content, spam, or misleading text into documents, but the effects are recoverable through undo/editing. Not destructive since the operation is reversible and doesn't permanently delete data.
From the tool's definition Tool name 'add_paragraph' combined with server description stating it 'enables AI agents to read, edit, and create Microsoft Word documents' and the sibling tools including write operations like 'add_heading', 'add_table', 'create_new_document', and…
Attacks that exploit this kind of access
Add a text paragraph. It is categorised as a Write tool in the Docx MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Docx MCP server in PolicyLayer and add a rule for add_paragraph: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Docx. Nothing to install.
add_paragraph is a Write tool with medium risk. Write tools should be rate-limited to prevent accidental bulk modifications.
Yes. Add a rate_limit block to the add_paragraph rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for add_paragraph. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
add_paragraph is provided by the Docx MCP server (okamifeng/docx-mcp-server). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →