Ask Questions of Your ClickHouse Data Through Your Muse

Your muse can ask questions of data stored in ClickHouse and answer in plain words, using a read-only connection that cannot change, add, or delete anything in the database.

What it does

ClickHouse is a database built for fast counts over very large tables, often used for website events. Through a connector with 10 read-only tools, your muse can list databases, list tables in one database with row counts and sizes, and describe a table column by column. It can search for a column by name across databases, check how much disk each table takes, see which queries are running now, and read the last 100 finished or failed queries from the past hour. For the answer itself, it can run one read-only query at a time, such as a count per day or a total by country. Every request carries a read-only setting, so ClickHouse itself refuses inserts, changes, and deletes. There are no write tools.

How to use it

1. Start an AnythingMCP workspace and install the ClickHouse connector.
2. Make a database user that can only read the tables you want to ask about, and put its server address, user name, and password into the connector.
3. On ClickHouse Cloud, add the address AnythingMCP calls from to the service IP access list, or the connection will be refused.
4. In Muse, open Settings, then Connectors, and add a custom connector with the MCP server address AnythingMCP gives you. Sign in and approve.
5. Check the setup by asking for the server version and its databases, then ask a question you can check.

What you need

• A Muse account that can add custom connectors
• A ClickHouse server or a ClickHouse Cloud service
• A read-only database user made for this connection
• An AnythingMCP account with the ClickHouse connector added

What to watch for

• Give this connection a user that can only read, and only the tables it needs. The read-only setting is a strong second lock, not a reason to hand over a full admin account.
• There is no row cap besides the limit in the query itself. Ask for totals and grouped results, not raw rows from a huge table, or the answer will be too large and slow.
• This connector does not manage a Cloud service. Starting, stopping, scaling, and backups need the separate Cloud connector or the Cloud console.
• This is a custom connector, and Meta does not review custom connectors. You can disconnect it in Settings at any time. The muse prepares, you decide.