Skip to main content
Follow these steps to run the ClickHouse MCP Server agent in a Meshagent Room for testing:
  1. Set Required Environment Variables
    Define these variables as needed for your ClickHouse connection. Replace the password token with your own secret.
    • CLICKHOUSE_HOST: sql-clickhouse.clickhouse.com
    • CLICKHOUSE_PORT: 8443
    • CLICKHOUSE_USER: demo
    • CLICKHOUSE_SECURE: true
    • CLICKHOUSE_VERIFY: true
    • CLICKHOUSE_CONNECT_TIMEOUT: 30
    • CLICKHOUSE_SEND_RECEIVE_TIMEOUT: 30
    • CLICKHOUSE_PASSWORD: <YOUR_CLICKHOUSE_PASSWORD_HERE>
  2. Install Meshagent 
    pip install "meshagent[all]"
  3. Sign Up & Authenticate
    Register and log in as shown in Meshagent CLI Getting Started.
  4. Start an MCP Server Agent in a Test Room
    Replace password and any other secrets accordingly:     
    meshagent service test --room=test --role=agent --image=meshagent/mcp-clickhouse:latest \
  --env MESHAGENT_PORT=8001 \
  --env CLICKHOUSE_HOST=sql-clickhouse.clickhouse.com \
  --env CLICKHOUSE_PORT=8443 \
  --env CLICKHOUSE_USER=demo \
  --env CLICKHOUSE_SECURE=true \
  --env CLICKHOUSE_VERIFY=true \
  --env CLICKHOUSE_CONNECT_TIMEOUT=30 \
  --env CLICKHOUSE_SEND_RECEIVE_TIMEOUT=30 \
  --env CLICKHOUSE_PASSWORD=`<YOUR_CLICKHOUSE_PASSWORD_HERE>` \
  --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" \
  --name=mcp-clickhouse-service-test
    • This starts a Meshagent Room with a test MCP server available. The agent and room are cleaned up automatically when inactive.
  5. Attach a Chatbot with ClickHouse Tools 
    meshagent chatbot join --room=test --agent-name=mcp-clickhouse --name=mcp-clickhouse --toolkit=mcp-clickhouse
    • This launches a chatbot in the room, ready to use the ClickHouse toolkit. You can load multiple toolkits per room/agent if needed.
    • A link to the new Room will appear in the output.
  6. Try It Out
    Open the provided room link in your browser and chat with the agent to run database queries and commands using the MCP Server tools.

​
Project Level Deployment

For production or persistent environments where you want MCP server tools and a chatbot automatically in every new MeshAgent room, deploy these as project-level services:
  1. Create Persistent MCP Server Service 
    meshagent service create --role=agent --image=meshagent/mcp-clickhouse:latest \
  --env MESHAGENT_PORT=8001 \
  --env CLICKHOUSE_HOST=sql-clickhouse.clickhouse.com \
  --env CLICKHOUSE_PORT=8443 \
  --env CLICKHOUSE_USER=demo \
  --env CLICKHOUSE_SECURE=true \
  --env CLICKHOUSE_VERIFY=true \
  --env CLICKHOUSE_CONNECT_TIMEOUT=30 \
  --env CLICKHOUSE_SEND_RECEIVE_TIMEOUT=30 \
  --env CLICKHOUSE_PASSWORD=`<YOUR_CLICKHOUSE_PASSWORD_HERE>` \
  --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" \
  --name=mcp-clickhouse-service
  2. Create Persistent Chatbot Service 
    meshagent service create --image="meshagent/cli:latest" \
  --port="num=9001 path=/agent liveness=/ type=meshagent.callable participant_name=mcp-clickhouse-chatbot" \
  --env="MESHAGENT_PORT=9001" \
  --name="mcp-clickhouse-chatbot-service" \
  --command="meshagent chatbot service --agent-name=mcp-clickhouse-chatbot --toolkit=mcp-clickhouse"
These services become available in every Meshagent Room created in the project. No need to run separate server or chatbot processes—they’ll join each new Room automatically.

​
Tools Available

Tools provided by the ClickHouse MCP Server for use in Meshagent Rooms:
ToolShort Description
list_databasesList available ClickHouse databases
list_tablesList available ClickHouse tables in a database, including schema, comment, row count, column count.
run_select_queryRun a SELECT query in a ClickHouse database

​
Tool Details

​
Tool: list_databases

  • Description: List available ClickHouse databases.

​
Tool: list_tables

  • Description: List available ClickHouse tables in a database, including schema, comment, row count, and column count.
  • Parameters:
    • database (string)
    • like (string, optional)
    • not_like (string, optional)

​
Tool: run_select_query

  • Description: Run a SELECT query in a ClickHouse database.
  • Parameters:
    • query (string)
⌘I