Gives Claude SQL access to CSV files, Postgres databases, and REST APIs through an embedded DuckDB engine. The standout feature is cross-source joins: you can write a single SELECT statement that joins a local CSV against a Postgres table and a REST endpoint response without shuffling data between tools. Exposes seven tools covering connection management, schema inspection, query execution, and health checks. Ships with a 1000-row result cap and read-only mode by default. REST responses are cached for five minutes to avoid redundant calls. If you're asking questions that span multiple data sources or want one server instead of three separate connectors, this replaces the usual per-source setup. Configuration lives in a YAML file with environment variable support for credentials.
claude mcp add --transport stdio dbsectrainer-mcp-data-pipeline-connector -- npx -y mcp-data-pipeline-connector