AI agents call download_dir to retrieve information from Nutstore without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool performs a data retrieval operation (download) from cloud storage to a local system. It has no side effects on the source data—files are not modified, deleted, or created in Nutstore. The action is non-destructive and read-only in nature. While downloading multiple files could have bandwidth/storage implications, this is an informational retrieval operation, not a write, execute, or destructive action.
From the tool's definition Tool description translates to: 'Recursively download an entire directory from Nutstore to local path (preserving directory structure)'. The operation is read-only, retrieving/downloading files without modifying or deleting data in Nutstore.
Attacks that exploit this kind of access
从坚果云递归下载整个目录到本机本地路径（保留目录结构）。. It is categorised as a Read tool in the Nutstore MCP Server, which means it retrieves data without modifying state.
Register the Nutstore MCP server in PolicyLayer and add a rule for download_dir: 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 Nutstore. Nothing to install.
download_dir 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_dir 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_dir. 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_dir is provided by the Nutstore MCP server (silverze/nutstore-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 →