AI agents call ftp_download to retrieve information from Ftp without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves a single file from a remote FTP/SFTP server with no side effects on the server state. It is a read operation that queries and transfers data. The severity is low because downloading files has minimal blast radius—no data is deleted, modified, or at risk of financial impact. The confidence is high given the clear descriptive language.
From the tool's definition Tool description states 'Standard single-file transport from the remote server' with recommendation to use batch_download for bulk operations. The verb 'download' and 'transport from remote server' indicate retrieval without modification.
Attacks that exploit this kind of access
Standard single-file transport from the remote server. For bulk downloads, favor ftp_batch_download. It is categorised as a Read tool in the Ftp MCP Server, which means it retrieves data without modifying state.
Register the Ftp MCP server in PolicyLayer and add a rule for ftp_download: 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 Ftp. Nothing to install.
ftp_download 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 ftp_download 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 ftp_download. 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.
ftp_download is provided by the Ftp MCP server (kynlos/ftp-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 →