Export Hugging Face dataset split as Parquet file
AI agents call get_parquet to retrieve information from Dataset Viewer MCP Server without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves and exports data in Parquet format without modifying the underlying dataset. Export/download operations are non-destructive reads that pose minimal security risk unless the exported data itself is sensitive, which depends on dataset contents rather than tool capability. No create, modify, delete, execute, or financial operations are involved.
From the tool's definition Tool description states 'Export Hugging Face dataset split as Parquet file' — exporting/downloading data is a read operation. The tool name 'get_parquet' also indicates retrieval rather than modification.
Documented attack patterns abuse exactly the kind of access get_parquet gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Dataset Viewer MCP Server, and nothing reaches the server without passing your rules. This is the rule we recommend for get_parquet:
{
"version": "1",
"default": "deny",
"tools": {
"get_parquet": {}
}
}get_parquet is read-only, so it stays allowed — but everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Export Hugging Face dataset split as Parquet file. It is categorised as a Read tool in the Dataset Viewer MCP Server MCP Server, which means it retrieves data without modifying state.
Register the Dataset Viewer MCP Server MCP server in PolicyLayer and add a rule for get_parquet: 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 Dataset Viewer MCP Server. Nothing to install.
get_parquet 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 get_parquet 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 get_parquet. 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.
get_parquet is provided by the Dataset Viewer MCP Server MCP server (privetin/dataset-viewer). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Deterministic rules across all 8 Dataset Viewer MCP Server tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.
Free to start. No card required.
8 Dataset Viewer MCP Server tools catalogued and risk-classified — across an index of 42,500+ MCP servers.