The query tool executes SELECT and nothing else, so the worst an assistant can do here is read something it should not have — a meaningfully smaller risk than a server that can also write. CData points at its paid products for full read-write access, so treat this build as the reporting path. The setup is the honest cost: a licensed JDBC driver, a Maven build and a properties file per source, which is a lot of ceremony before the first question, and near-zero after.
A local, read-only MCP server that puts Act CRM behind a relational SQL model. It wraps CData's JDBC driver for Act CRM, which exposes the CRM's objects as tables, and lets an assistant discover those tables and run SELECT queries against live data in response to plain-language questions.
- The tables available in Act CRM, returned as CSV with a header row — `get_tables`
- The columns on a chosen table, same CSV form, so a query can be written against real column names — `get_columns`
- A SQL SELECT executed against live Act CRM data — `run_query`
- Tool names prefixed with the server name you set in the properties file, so several CData servers can sit side by side in one client without colliding
- A `Tables` property that limits which tables are reachable at all — leave it blank for everything, or list the ones you want exposed
An Act CRM account, plus CData's JDBC driver for Act CRM downloaded and licensed separately — running `java -jar cdata.jdbc.actcrm.jar --license` takes your name, email and either TRIAL or a license key. Java, and Maven to build: `mvn clean install` produces the runnable JAR. Then write a `.prp` file naming the prefix, server name and version, the path to the driver JAR, the driver class `cdata.jdbc.actcrm.ActCRMDriver`, and the JDBC connection string produced by the driver's connection-string utility — for OAuth sources that utility opens a browser to authenticate. The client entry runs java with the JAR and the `.prp` path as arguments. The server communicates over stdio, so the client must be on the same machine. MIT licensed.
Build from source — clone the repository and build it, then deploy it and point your client at the endpoint
