AI agents use edit_file to create or update resources in Overleaf — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Overleaf environment.
This tool creates or modifies data (LaTeX document contents) through a reversible operation. While editing files can have downstream effects (e.g., breaking a document), the action itself is not destructive—changes can be undone via version control or Overleaf's undo functionality. The medium severity reflects that a misguided agent could corrupt academic work or collaborative documents, but damage is recoverable.
From the tool's definition Tool description explicitly states it 'Replaces the contents of a doc' and 'submitting it as an OT operation', indicating it modifies file contents reversibly through Overleaf's API.
Attacks that exploit this kind of access
Replaces the contents of a doc by computing a minimal diff and submitting it as an OT operation. It is categorised as a Write tool in the Overleaf MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Overleaf MCP server in PolicyLayer and add a rule for edit_file: 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 Overleaf. Nothing to install.
edit_file 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 edit_file 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 edit_file. 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.
edit_file is provided by the Overleaf MCP server (netique/overleaf-mcp). 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 →