Get up and running with the MCP Server and tools in a cloud-based Room for quick, disposable experiments or development.

  1. Install Meshagent CLI

    pip install "meshagent[all]"

  2. Sign up and Authenticate

  3. Start the MCP Server Agent in a Room

    meshagent service test --room=test --role=agent --image=meshagent/mcp-keboola-mcp:latest --env MESHAGENT_PORT=8001 --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" --name=mcp-keboola-mcp-service-test
    • This will provision a temporary Room called test and launch the MCP server as a cloud agent inside. When the Room is closed (e.g., participants leave), both the MCP server and the agent are torn down automatically.

  4. Start a Chatbot with MCP Tools in the Room

    meshagent chatbot join --room=test --agent-name=mcp-keboola-mcp --name=mcp-keboola-mcp --toolkit=mcp-keboola-mcp
    • This starts a chatbot in the Room, able to invoke all MCP tools. You can use multiple toolkits or agents in the same Room.
    • The output provides a link to the Room UI in your browser.

  5. Interact

    • Visit the provided Room link and send a message to the agent to interact with the MCP Server tools in real time.

​
Project Level Deployment (Persistent Services)

For production and persistent environments, you can deploy MCP Server tools and agents at the Meshagent project level. This ensures the tools and chatbot are automatically available in all Rooms within the project, without manual startup or teardown.

  1. Create Persistent MCP Server Agent Service

    meshagent service create --role=agent --image=meshagent/mcp-keboola-mcp:latest --env MESHAGENT_PORT=8001 --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" --name=mcp-keboola-mcp-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-keboola-mcp-chatbot" --env="MESHAGENT_PORT=9001" --name="mcp-keboola-mcp-chatbot-service" --command="meshagent chatbot service --agent-name=mcp-keboola-mcp-chatbot --toolkit=mcp-keboola-mcp"

Once configured, the MCP server agent and chatbot are automatically available and running every time a Room is spun up in your Meshagent project.

​
Tools Available

The following tools are exposed by the MCP Server agent, available for use by Meshagent chatbots or remote control:

​
Tool Summary

Tool NameShort Description
add_config_rowCreates a component configuration row in the specified configuration_id, using the specified name, component ID, configuration JSON, and description.
create_configCreates a root component configuration using the specified name, component ID, configuration JSON, and description.
create_flowCreates a new flow configuration in Keboola.
create_sql_transformationCreates an SQL transformation using the specified name, SQL query, detailed description, and created table names.
docs_queryAnswers a question using the Keboola documentation as a source.
find_component_idReturns list of component IDs that match the given query.
get_bucketGets detailed information about a specific bucket.
get_componentGets information about a specific component given its ID.
get_configGets information about a specific component/transformation configuration.
get_config_examplesRetrieves sample configuration examples for a specific component.
get_flowGets detailed information about a specific flow configuration.
get_flow_schemaReturns the JSON schema that defines Flow structure.
get_jobRetrieves information about a specific job including status and metadata.
get_project_infoReturns structured project information from multiple endpoints.
get_sql_dialectGets the name of the SQL dialect used by Keboola’s database.
get_tableGets detailed information about a specific table.
list_bucketsRetrieves all buckets in the project.
list_configsRetrieves component configurations in the project (filterable).
list_flowsRetrieves flow configurations from the project.
list_jobsRetrieves all jobs (filterable by component ID, status, pagination, etc).
list_tablesRetrieves all tables in a specific bucket.
list_transformationsRetrieves transformation configurations in the project (filterable).
query_dataExecutes an SQL SELECT query on Keboola database and returns data.
run_jobStarts a new job for a given component or transformation.
update_bucket_descriptionUpdates a bucket’s description.
update_column_descriptionUpdates a table column’s description.
update_configUpdates a root component configuration with provided parameters.
update_config_rowUpdates a component configuration row with provided parameters.
update_flowUpdates an existing flow configuration.
update_sql_transformationUpdates an existing SQL transformation configuration.
update_table_descriptionUpdates the description for a given table.

​
Tool Details

Get up and running with the MCP Server and tools in a cloud-based Room for quick, disposable experiments or development.

  1. Install Meshagent CLI

    pip install "meshagent[all]"

  2. Sign up and Authenticate

  3. Start the MCP Server Agent in a Room

    meshagent service test --room=test --role=agent --image=meshagent/mcp-keboola-mcp:latest --env MESHAGENT_PORT=8001 --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" --name=mcp-keboola-mcp-service-test
    • This will provision a temporary Room called test and launch the MCP server as a cloud agent inside. When the Room is closed (e.g., participants leave), both the MCP server and the agent are torn down automatically.

  4. Start a Chatbot with MCP Tools in the Room

    meshagent chatbot join --room=test --agent-name=mcp-keboola-mcp --name=mcp-keboola-mcp --toolkit=mcp-keboola-mcp
    • This starts a chatbot in the Room, able to invoke all MCP tools. You can use multiple toolkits or agents in the same Room.
    • The output provides a link to the Room UI in your browser.

  5. Interact

    • Visit the provided Room link and send a message to the agent to interact with the MCP Server tools in real time.

​
Project Level Deployment (Persistent Services)

For production and persistent environments, you can deploy MCP Server tools and agents at the Meshagent project level. This ensures the tools and chatbot are automatically available in all Rooms within the project, without manual startup or teardown.

  1. Create Persistent MCP Server Agent Service

    meshagent service create --role=agent --image=meshagent/mcp-keboola-mcp:latest --env MESHAGENT_PORT=8001 --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" --name=mcp-keboola-mcp-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-keboola-mcp-chatbot" --env="MESHAGENT_PORT=9001" --name="mcp-keboola-mcp-chatbot-service" --command="meshagent chatbot service --agent-name=mcp-keboola-mcp-chatbot --toolkit=mcp-keboola-mcp"

Once configured, the MCP server agent and chatbot are automatically available and running every time a Room is spun up in your Meshagent project.

​
Tools Available

The following tools are exposed by the MCP Server agent, available for use by Meshagent chatbots or remote control:

​
Tool Summary

Tool NameShort Description
add_config_rowCreates a component configuration row in the specified configuration_id, using the specified name, component ID, configuration JSON, and description.
create_configCreates a root component configuration using the specified name, component ID, configuration JSON, and description.
create_flowCreates a new flow configuration in Keboola.
create_sql_transformationCreates an SQL transformation using the specified name, SQL query, detailed description, and created table names.
docs_queryAnswers a question using the Keboola documentation as a source.
find_component_idReturns list of component IDs that match the given query.
get_bucketGets detailed information about a specific bucket.
get_componentGets information about a specific component given its ID.
get_configGets information about a specific component/transformation configuration.
get_config_examplesRetrieves sample configuration examples for a specific component.
get_flowGets detailed information about a specific flow configuration.
get_flow_schemaReturns the JSON schema that defines Flow structure.
get_jobRetrieves information about a specific job including status and metadata.
get_project_infoReturns structured project information from multiple endpoints.
get_sql_dialectGets the name of the SQL dialect used by Keboola’s database.
get_tableGets detailed information about a specific table.
list_bucketsRetrieves all buckets in the project.
list_configsRetrieves component configurations in the project (filterable).
list_flowsRetrieves flow configurations from the project.
list_jobsRetrieves all jobs (filterable by component ID, status, pagination, etc).
list_tablesRetrieves all tables in a specific bucket.
list_transformationsRetrieves transformation configurations in the project (filterable).
query_dataExecutes an SQL SELECT query on Keboola database and returns data.
run_jobStarts a new job for a given component or transformation.
update_bucket_descriptionUpdates a bucket’s description.
update_column_descriptionUpdates a table column’s description.
update_configUpdates a root component configuration with provided parameters.
update_config_rowUpdates a component configuration row with provided parameters.
update_flowUpdates an existing flow configuration.
update_sql_transformationUpdates an existing SQL transformation configuration.
update_table_descriptionUpdates the description for a given table.

​
Tool Details