Access Sigma from an AI assistant, agent, or tool

Sigma supports multiple ways to connect your Sigma organization to AI assistants, coding agents, and chat interfaces. After connecting an AI assistant, you can view, query, edit, and manage content in Sigma with the help of your preferred natural language tool.

This document covers the methods available to connect your organization to AI tools, the features and resources that support your agents in Sigma, and example use cases.

Connect Sigma to an AI assistant

Connect an AI assistant to your Sigma organization so that it can perform analysis, take action, and create new workbooks, reports, and data models.

You can give an AI assistant access to Sigma in the following ways:

Sigma MCP server

Connect your AI assistant to the Sigma MCP server so it can search documents and query data using the same permissions you have in your Sigma organization.

Sigma plugin

Use the Sigma plugin with a supported tool to perform complex analysis and build workbooks in an agentic workflow.

Sigma REST API

Give an AI assistant access to the Sigma REST API to programmatically administer your organization and build documents from code.

Sigma CLI

Configure the Sigma CLI to give agents simple, structured commands for the Sigma REST API.

Resources for AI assistants in Sigma

After connecting your AI assistant to Sigma, direct it to the following resources and features to help it act and build in Sigma:

Agent skills for Sigma

Install agent skills for Sigma to help your AI assistants use the Sigma REST API, install the Sigma CLI, and write code representations of documents.

Migration skills for Sigma (Beta)

Use migration skills to provide reference materials and scripts your assistant can use to translate content from Tableau, Power BI, Qlik, ThoughtSpot, and other BI tools to Sigma.

OpenAPI specification for the Sigma REST API

Give agents direct access to the source of truth for parameters, requests, and responses in the Sigma REST API. Includes the schemas for creating Sigma documents as code.

Manage data models as code

Read, write, and update data models based on a code representation. Give AI assistants an agent-friendly version of any Sigma data model to analyze, edit, and build.

Manage workbooks as code (Beta)

Read, write, and update workbooks based on a code representation. Give AI assistants an agent-friendly version of any Sigma workbook to analyze, edit, and build.

Manage reports as code (Beta)

Read, write, and update reports based on a code representation. Give AI assistants an agent-friendly version of any Sigma report to analyze, edit, and build.

Sigma documentation

Give agents access to the official Sigma documentation using the documentation MCP server, along with an llms.txt index and agent-ready versions of all pages served in Markdown.

Decide how to connect your AI assistant to Sigma

When deciding how to connect your AI assistant to Sigma, consider the tradeoffs for each method.

Method	Best for	Tradeoffs
MCP server	Asking natural language questions about your data with your existing Sigma account	Can’t create or edit documents in Sigma
Plugin	Analyzing data and building workbooks directly from a supported AI assistant	Available only for specific AI assistants
REST API	Full programmatic access, including creating documents and administering resources	Requires working knowledge of the Sigma REST API
CLI	REST API access through an intuitive command-line interface	Requires installation and setup

The REST API and CLI can give agents access to delete content, export data, and perform other sensitive operations. Configure permissions carefully before connecting an agent.

Example agent workflows with Sigma

After you connect your AI assistant to Sigma and give it resources and reference materials, you can describe tasks in natural language and have the assistant analyze data and create documents. Consider the example configurations below for working agentically with Sigma.

Analyze data with the Sigma MCP server

Analyze data from Sigma with the AI assistant you use every day

Confirm you have access

Start by making sure your account type has the Use Sigma MCP with OAuth permission enabled. If it isn’t, ask an organization admin to enable it.

Connect your AI assistant to the MCP server

Follow the steps in Use the Sigma MCP server to configure the Sigma MCP server with your preferred AI assistant, such as Claude, ChatGPT, Codex, Cortex Code, or Cursor.

Start analyzing

After successfully authenticating to the Sigma MCP server, start asking questions about data in your Sigma organization. For suggestions, see Usage examples.

Build workbooks from code with agent skills and the Sigma CLI

Set up your agent to read, manage, and build workbooks autonomously

Confirm you have access

Start by making sure your account type has all the right permissions, and that you meet all the user and system requirements for using the Sigma CLI, installing agent skills for Sigma, and managing workbooks as code.

Install agent skills for Sigma

Follow the steps to install agent skills for Sigma with the AI assistant of your choice. Agent skills can be used to install and use the Sigma CLI, as well as provide reference materials that support agents authoring documents as code in Sigma.

Set up the Sigma CLI

Install and configure the Sigma CLI or use the sigma-cli skill installed in step two to help you install it. The Sigma CLI allows you to query the Sigma REST API using structured commands, as well as authenticate via OAuth to your Sigma account. This makes it easier for agents to handle authentication, learn about your data, and take action.

Ask the agent to create a new workbook

Prompt your AI assistant to create a new workbook for you in Sigma, whether it’s a new version of an existing dashboard or an entirely new workflow. Using the sigma-workbooks skill you installed in step two, and the Sigma CLI, the agent can read reference materials about authoring workbooks as code and query existing content in your Sigma organization to build new documents from scratch.

Continue agentic work in Sigma

Have your AI assistant hand off agentic work to a governed Sigma agent

Build a Sigma agent

For agentic workflows that require role-based access control and data security, start by building a Sigma agent and configuring it with data sources, instructions, and actions. This allows your AI assistant to hand off work to a governed Sigma agent that can take action on your behalf without removing any of the security or governance controls in your Sigma organization.

Connect your AI assistant to the Sigma REST API

Set up your AI assistant to connect to the Sigma REST API using OAuth or an API key. Consider installing agent skills for Sigma so your assistant can use the sigma-api skill to learn about the Sigma REST API.

Call the Run a Sigma agent endpoint

Once your AI assistant is connected to the Sigma REST API, you can call the Run a Sigma agent endpoint to execute the agent’s actions. This allows your AI assistant to interact with the Sigma agent and trigger specific workflows or tasks. It can receive a response directly from the Sigma agent and continue working with the results, or it can hand off the work to the Sigma agent and let it complete the task in Sigma.