AI agents call download_attachment to retrieve information from Gmail without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
The tool retrieves/downloads existing data (an attachment) without altering, creating, or destroying it. While it writes the attachment to a local location, this is a standard side effect of retrieval operations and does not constitute data modification in the system. No financial impact, no code execution, and no irreversible changes.
From the tool's definition Tool name 'download_attachment' and description 'Downloads an email attachment to a specified location' indicates retrieval of data with no modification or deletion of the attachment or email itself.
Attacks that exploit this kind of access
Downloads an email attachment to a specified location. It is categorised as a Read tool in the Gmail MCP Server, which means it retrieves data without modifying state.
Register the Gmail MCP server in PolicyLayer and add a rule for download_attachment: 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 Gmail. Nothing to install.
download_attachment is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the download_attachment 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 download_attachment. 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.
download_attachment is provided by the Gmail MCP server (pouyanafisi/gmail-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 →