Tako MCP

What is Tako MCP about?

Tako MCP enables you to query the Tako platform for up‑to‑date information and instantly turn that data into visualizations (embeds, webpages, images) via Model Context Protocol tools.

How to use Tako MCP?

1. Obtain a Tako API key from the Tako Dashboard.
2. Install the server (e.g., via Smithery CLI) and configure the environment variable TAKO_API_KEY.
3. Run the server using the provided command (e.g., uv run main.py).
4. In a Claude Desktop or other MCP‑compatible client, add the server configuration to mcp.json.
5. Call the exposed tools such as search_tako, upload_file_to_visualize, visualize_file, or visualize_dataset from your prompts.

Key Features of Tako MCP

• search_tako: Query Tako and the web for real‑time data, returning embeds, webpages, and image URLs with metadata.
• upload_file_to_visualize: Upload a base64‑encoded file to Tako for later visualization.
• visualize_file: Produce visual output from a previously uploaded file.
• visualize_dataset: Directly visualize data formatted in the Tako Data Format.
• Generated Prompts: Helper prompts (generate_search_tako_prompt, generate_visualization_prompt) assist users in crafting optimal queries and visualization requests.
• Remote & Local Modes: Switch between ENVIRONMENT=remote and ENVIRONMENT=local to run the server where needed.

Use Cases of Tako MCP

• Research Reports: Quickly gather and visualize financial metrics for a set of companies.
• Data Dashboards: Embed live visualizations in internal tools or web pages.
• Educational Content: Produce illustrated explanations of statistical concepts on the fly.
• Business Intelligence: Retrieve market data and generate charts for decision‑making meetings.

FAQ

Q: Where do I get the Tako API key? A: Sign in to the Tako Dashboard (https://trytako.com/dashboard) and copy the key from the API section.

Q: Do I need a specific Python environment? A: The project uses uv for dependency management; running uv sync will set up required packages.

Q: How do I run the server in remote mode? A: Set the environment variable ENVIRONMENT=remote before starting the server, e.g., ENVIRONMENT=remote TAKO_API_KEY=… uv run main.py.

Q: Can I test the tools without writing code? A: Yes, use the Model Context Protocol Inspector (npx @modelcontextprotocol/inspector@latest) to connect to the running MCP and invoke tools interactively.

Q: What if I need to visualize large files? A: Use the upload_file_to_visualize tool sparingly; for large files, the README recommends the Tako Playground (https://trytako.com/playground).

Tako MCP

Tako MCP is a simple MCP server that queries Tako and returns real-time data and visualization

Check out Tako and our documentation

Available Tools

search_tako

Takes a query to search Tako and the web to get real-time data and visualization. Returns embed, webpage, and image url of the visualization with relevant metadata such as source, methodology, and description.

upload_file_to_visualize

Takes a base64 encoded file as an input and uploads it to Tako to use for visualization

*If you call this tool with a big file, it may consume a large number of tokens and will be very slow. If you want to test visualizing bigger files though Tako, visit our playground

visualize_file

Use the file_id from upload_file_to_visualize and visualize the file. Returns embed, webpage, and image url of the visualization

visualize_dataset

Takes a Tako Data Format data and visualize. Returns embed, webpage, and image url of the visualization

Available Prompts

generate_search_tako_prompt

Prompt to assist the client to format query and search Tako using search_tako tool

generate_visualization_prompt

Prompt to assist the client to transform the data into Tako Data Format and visualize using visualize_dataset tool

Quickstart

Get your API key

Access Tako Dashboard and get your API key.

Installing via Smithery

To install tako-mcp for Claude Desktop automatically via Smithery:

npx -y @smithery/cli install @TakoData/tako-mcp --client claude

Add Tako MCP to Claude Desktop

Add the following to your .cursor/mcp.json or claude_desktop_config.json (MacOS: ~/Library/Application\ Support/Claude/claude_desktop_config.json)

{
    "mcpServers": {
        "takoApi": {
            "command": "uv",
            "args": [
                "--directory",
                "/path/to/tako/mcp",
                "run",
                "main.py"
            ],
            "env": {
                "TAKO_API_KEY": "<TAKO_API_KEY>"
            }
        }
    }
}

Example:

1. Use the prompt from Tako MCP Server generate_search_tako_prompt

The prompt will guide the model to generate optimized query to search Tako

2. Add your text input

Add an input text to generate the prompt

"Compare Magnificent 7 stock companies on relevant metrics."

3. Add a prompt to the chat

Add additional instructions to the chat prompt

Write me a research report on the magnificent 7 companies. Embed the result in an iframe whenever necessary

4. Checkout the result

• Claude Response
• Generated Report

Environment Variables

ENVIRONMENT

Options:

• remote - If you're running a remote MCP server
• local - If you're running a local MCP server

TAKO_API_KEY

• Your Tako API key, access it from Tako Dashboard

Testing Remote MCP

Start inspector and access the console

npx -y npx @modelcontextprotocol/inspector@latest

Start Tako MCP Server on remote mode

ENVIRONMENT=remote TAKO_API_KEY=<your_tako_api_key> uv run main.py

In inspector console, add the url https://0.0.0.0:<port>/mcp/ and click connect

Select the Tools tab, and click ListTools.

Select search_tako and test a query

Deploying it on render

Since we use uv Render uses pip, we have to build a requirements.txt

uv pip compile pyproject.toml > requirements.txt