rudof is a Rust library that offers Semantic Web operations (RDF processing, SPARQL queries, ShEx/SHACL validation, among others). The Model Context Protocol (MCP) provides a standardized interface for AI agents to interact with external systems. This demo presents rudof-MCP, an MCP server exposing rudof's functionalities to AI agents through natural language. By delegating Semantic Web operations to rudof at inference time, LLMs can reduce dependence on pretraining knowledge and ground their outputs in auditable tool invocations that ensure the accuracy and reproducibility of results. In this demo, we showcase this by comparing Claude with and without access to rudof-MCP on a set of Semantic Web tasks.