Get up and running with the MCP Vizro server in a Meshagent Room, and interact with its tools via an agent in minutes:

  1. Install the Meshagent Python SDK:

    pip install "meshagent[all]"

  2. Sign up and authenticate:

  3. Start the Vizro MCP service in a room:

    meshagent service test --room=test --role=agent --image=meshagent/mcp-vizro:latest --env MESHAGENT_PORT=8001 --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" --name=mcp-vizro-service
    • This command launches a Meshagent Room named test, hosting the Vizro MCP server toolkit as a callable service.
    • Note: Rooms close automatically if they go inactive.

  4. Join the room with a chatbot agent using the Vizro toolkit:

    meshagent chatbot join --room=test --agent-name=sample --name=sample --toolkit=mcp-vizro
    • This starts a chatbot in the Meshagent Room, with access to the full mcp-vizro toolkit.
    • Multiple toolkits can be added to the same room/agent, enabling complex workflows.
    • The output of this command will provide a link to the active Meshagent Room, where you can interact with your agent and the Vizro tools via chat.

​
Tools Available

Below are the tools exposed by the Vizro MCP Server for use within Meshagent Rooms.

​
Tools Summary

Tool NameShort Description
get_model_json_schemaGet the JSON schema for the specified Vizro model.
get_sample_data_infoRetrieve information about available sample datasets.
get_vizro_chart_or_dashboard_planGet instructions for creating a Vizro chart or dashboard.
load_and_analyze_dataLoad and analyze data from various file formats.
validate_chart_codeValidate and (optionally) preview Vizro chart code.
validate_model_configValidate a Vizro dashboard model configuration.

​
Tool Details

​
get_model_json_schema

Get the JSON schema for the specified Vizro model.

ParameterTypeDescription
model_namestringName of the Vizro model to get schema for (e.g., ‘Card’, ‘Dashboard’, ‘Page’)

​
get_sample_data_info

Retrieve information about sample datasets for prototyping and exploration.

Use the following data for the below purposes:

  • iris: Mostly numerical with one categorical column — good for scatter, histogram, boxplot, etc.
  • tips: Mixed numerical and categorical columns — good for bar, pie, etc.
  • stocks: Stock prices — good for line, scatter, or time-series charts.
  • gapminder: Demographic data — good for line, scatter, maps, or multi-category plots.
ParameterTypeDescription
data_namestringName of the dataset to retrieve information about

​
get_vizro_chart_or_dashboard_plan

Get step-by-step instructions for constructing a Vizro chart or dashboard.

ParameterTypeDescription
user_planstringUser-provided plan or goal specification

​
load_and_analyze_data

Load data from local files or URLs into a pandas DataFrame and obtain metadata about its structure.

Supported formats:

  • CSV (.csv)
  • JSON (.json)
  • HTML (.html, .htm)
  • Excel (.xls, .xlsx)
  • OpenDocument Spreadsheet (.ods)
  • Parquet (.parquet)
ParameterTypeDescription
path_or_urlstringPath or URL to the data file

​
validate_chart_code

Validate generated chart code and optionally open a PyCafe visualization link in a browser.

ParameterTypeDescription
chart_configstringObject describing the chart configuration (ChartPlan)
data_infostringMetadata about the dataset to use in the chart
auto_openboolean*(Optional) Whether to automatically preview the chart in a browser

​
validate_model_config

Validate Vizro dashboard configuration and get Python code or a shareable link.

If successful, returns the valid Python code for your dashboard, and if applicable, a PyCafe link which may be automatically opened in your browser.

ParameterTypeDescription
dashboard_configobject(JSON string or dict) Vizro dashboard model configuration
data_infosarrayList of DFMetaData objects with data file info
auto_openboolean*(Optional) Whether to auto-open the PyCafe link in a browser