Tako

You can run Tako's MCP server to access Tako's knowledge base and interactive charts through a simple protocol. This server lets your agents search charts, fetch previews and insights, and render interactive Tako visualizations via MCP-UI, enabling powerful data-driven interactions with Tako's data assets.

How to use

Install the MCP server on your machine or in your environment, then start it and point your MCP client to the server URL. Use the available tools to search for charts, obtain chart previews, fetch AI-generated insights, and open interactive visualizations in MCP-UI.

How to install

Prerequisites: Python and a functioning internet connection. You also need a valid Tako API token to access Tako’s data.

Install the MCP server package from PyPI or install from source to run the server locally.

pip install tako-mcp

If you prefer to install from source, clone the repository and install in editable mode.

git clone https://github.com/anthropics/tako-mcp.git
cd tako-mcp
pip install -e .

Run the server directly or use Docker to containerize the service.

tako-mcp

docker build -t TakoMCP .
docker run -p 8001:8001 TakoMCP

Connect your MCP client to the server using this URL by default: http://localhost:8001.

Configuration and runtime details

Environment variables configure how the server reaches Tako and how it exposes its API. The following environment variables are used by the server and should be supplied as needed.

TAKO_API_URL: Tako API endpoint (default: https://api.trytako.com)
PUBLIC_BASE_URL: Public URL for chart embeds (default: https://trytako.com)
PORT: Server port (default: 8001)
HOST: Server host (default: 0.0.0.0)
MCP_ALLOWED_HOSTS: Additional allowed hosts (comma-separated)
MCP_ENABLE_DNS_REBINDING: Enable DNS rebinding protection (default: true)

Security and health checks

Use a secret API token from your Tako account to authorize requests. The server provides health endpoints to verify availability and readiness.

Troubleshooting tips

If you cannot reach the MCP server, verify network access to the host and port, ensure the API token is valid, and confirm the server process is running. Check logs for startup messages and errors related to API connectivity.

Example workflow with an MCP client

1. Start the MCP server on your host. 2. Point your MCP client to the server URL. 3. Use the knowledge search tool to find charts relevant to your topic. 4. Retrieve chart previews and AI-generated insights for chosen charts. 5. Open interactive charts in MCP-UI to explore data dynamically.

Notes on MCP-UI integration

An interactive chart can be rendered as an MCP-UI resource in compatible clients. The UI supports zooming, panning, hover interactions, and responsive resizing via postMessage. It adapts to light and dark themes to fit your environment.

Start commands and container usage

Direct run command to start the server locally.

tako-mcp

Containerized start command to run via Docker.

docker build -t TakoMCP .
docker run -p 8001:8001 TakoMCP

Available tools

knowledge_search

Search Tako's knowledge base for charts and data visualizations.

get_chart_image

Retrieve a preview image URL for a chart.

get_card_insights

Obtain AI-generated insights and a natural language summary for a chart.

explore_knowledge_graph

Discover entities, metrics, and cohorts within Tako's data graph.

list_chart_schemas

List available chart templates for ThinViz charts.

get_chart_schema

Get detailed information about a specific chart template.

create_chart

Create a chart from a template using your data.

open_chart_ui

Open an interactive chart in the MCP-UI for live exploration.