AI agents call create_dataset to retrieve information from Langsmith without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
Despite the name suggesting a Write operation, the description explicitly states this is a documentation-only tool, meaning it only retrieves/displays information about how to create datasets. No actual data creation occurs. Classified as Read with low severity since it only surfaces documentation.
From the tool's definition 'This is a documentation-only tool.' - the tool only provides documentation/information about how to create datasets, it does not actually create anything.
Attacks that exploit this kind of access
Call this tool when you need to understand how to create datasets in LangSmith. This is a documentation-only tool. It is categorised as a Read tool in the Langsmith MCP Server, which means it retrieves data without modifying state.
Register the Langsmith MCP server in PolicyLayer and add a rule for create_dataset: 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 Langsmith. Nothing to install.
create_dataset 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 create_dataset 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 create_dataset. 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.
create_dataset is provided by the Langsmith MCP server (langsmith-mcp-server). 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 →