Advertising APIs are awkward to query directly and usually end up copied into a warehouse first. Modelling the account as tables means a question about campaign performance is a SELECT rather than a pipeline, and read-only execution means pointing it at a live account cannot change a bid. CData names the alternatives openly: their downloadable MCP server for write access and a simpler setup, and Connect AI for a managed remote endpoint.
A local, read-only MCP server from CData wrapping their JDBC driver for Bing Ads. The driver presents the advertising account as relational SQL models; the server gives a model three tools over it, so campaign and performance questions can be answered without anyone writing SQL.
- A tool that lists the tables available in the data source, returned as CSV with a header row
- A tool that lists the columns on a named table, so field names get checked rather than guessed
- A tool that executes a SQL SELECT query against Bing Ads
- Scope control through the `Tables` property in the connection file: blank exposes everything, or name explicitly what should be reachable
- Tool names carry the prefix you set, so several CData servers can run side by side without colliding
The CData JDBC Driver for Bing Ads, downloaded and licensed separately — the driver jar takes your name, email and either TRIAL or a licence key. Java to run and Maven to build: `mvn clean install` produces `CDataMCP-jar-with-dependencies.jar`. Then a `.prp` file with `Prefix=bingads`, a server name and version, the driver path, `DriverClass=cdata.jdbc.bingads.BingAdsDriver`, the `JdbcUrl` built with the driver's connection-string utility, and `Tables`. Bing Ads authenticates over OAuth, so expect a browser step while building the connection string. The client launches java with the jar and the .prp path; transport is stdio, so client and server share a machine. MIT licensed.
Build from source — clone the repository and build it, then deploy it and point your client at the endpoint