Migrate to Sigma with an AI assistant (Beta)

This documentation describes one or more public beta features that are in development. Beta features are subject to quick, iterative changes; therefore the current user experience in the Sigma service can differ from the information provided in this page.

This page should not be considered official published documentation until Sigma removes this notice and the beta flag on the corresponding feature(s) in the Sigma service. For the full beta feature disclaimer, see Beta features.

You can use migration skills for Sigma to have an AI assistant rebuild your source content — the dashboards, reports, and data models in another business intelligence (BI) tool — as a Sigma document. Each supported source tool has its own skill, and most also have a read-only assessment skill that inventories your existing content so you can decide what to migrate first.

Migration skills allow you to use AI assistants to reduce the time it takes to migrate content to Sigma. Instead of manually reconstructing and verifying your data models by hand, the skills allow an AI assistant to rebuild and verify the content for you. After building the data model, the assistant uses the Sigma MCP server to query it, and compares the results against the same data platform your source content reads from. The migration isn’t complete until the values match.

Sigma publishes these migration skills as open source under the Apache 2.0 License. Review the migration skills notice before installing or using these migration skills.

Requirements

You need an AI assistant that supports skills, such as Claude Code, Cursor, Cortex Code, or Codex.
You must have client credentials for the Sigma REST API and the API base URL for your organization, or authenticate to the Sigma CLI with OAuth.
You must have the Sigma MCP server connected in your AI assistant.
You must have a Sigma connection to the same data platform your source content queries. Snowflake, Databricks, BigQuery, Redshift, and PostgreSQL connections are supported.
You must have access to your source tool. Each skill documents the credentials it needs, such as a Tableau personal access token or a Qlik API key.
You must have Ruby, Python 3, Node.js 18 or later, and Bash installed on your machine.

Limitations

Migration skills compare results against your data platform, so the source content must read from a data platform that Sigma can connect to. If the source data exists only in an extract or an imported model, you must load it into your data platform before you migrate.
The migration skills use the data model code representation and workbook code representation endpoints to recreate Sigma documents. Any limitations that apply to code representation apply to migration skills as well. See Manage data models as code and Manage workbooks as code.

Supported source tools for migration

Skills are available to help AI assistants migrate from the following source tools to Sigma:

Source tool	Skill	What it converts
Tableau	`tableau-to-sigma`	Data sources and workbooks
Power BI	`powerbi-to-sigma`	Reports and semantic models, including DAX measures
Qlik Sense and Qlik Cloud	`qlik-to-sigma`	Apps, including expressions and set analysis
ThoughtSpot	`thoughtspot-to-sigma`	Models, worksheets, and Liveboards
Amazon QuickSight	`quicksight-to-sigma`	Analyses and dashboards
IBM Cognos	`cognos-to-sigma`	Data modules and report specifications
Looker	`looker-to-sigma`	LookML models and dashboards
MicroStrategy	`microstrategy-to-sigma`	Dossiers, reports, and schemas
Sisense	`sisense-to-sigma`	ElastiCube and Live models, and dashboards
GoodData	`gooddata-to-sigma`	Workspaces, including logical data models, MAQL metrics, and insights
Domo	`domo-to-sigma`	Datasets, Beast Mode calculations, and cards

For many source tools, Sigma provides an assessment skill, named <source-tool>-assessment, that inventories your existing content and ranks it by complexity. Some source tools have additional skills that move data into your data platform before a migration. For details, see the documentation in the sigma-migration-skills repository.

How the migration skills perform migrations

During a migration, the skills go through the following phases and stop when the AI assistant needs you to make a decision or provide input:

Phase	What happens
Assess	Inventories your source content and ranks it by complexity. This phase is read-only.
Discover	Reads the source model, dashboards, and data platform columns.
Convert	Translates source measures, calculations, and expressions into Sigma formulas.
Data model	Builds a Sigma data model with tables, relationships, and metrics.
Build	Rebuilds the source pages and visualizations as a Sigma workbook.
Layout	Applies the workbook layout.
Verify	Uses the Sigma MCP server to query the new Sigma document, queries the source data platform, and compares the results.
Security	Reports any row-level or column-level security found in the source, then applies it after you approve.

Install the migration skills

Agent skills for migration to Sigma are distributed through the sigma-migration-skills repository.

Install only the skills for the source tools you’re migrating from, plus the sigma-authoring skill. This skill provides the general Sigma reference materials that each source tool’s skill relies on.

For example, to install the migration skills as plugins for Claude Code, run the following commands:

/plugin marketplace add https://github.com/sigmacomputing/sigma-migration-skills.git
/plugin install sigma-authoring@sigma-migration-skills
/plugin install <source-tool>-to-sigma@sigma-migration-skills

To use migration skills with another AI assistant, clone the repository and point your assistant at the skill directories you need:

git clone https://github.com/sigmacomputing/sigma-migration-skills.git

For installation steps for other AI assistants, see Install skills for AI assistants.

Configure your Sigma credentials

Before your first migration, set the following environment variables so the skills can authenticate to the Sigma REST API:

# The API base URL for your Sigma organization.
export SIGMA_BASE_URL='<api-base-url>'
# Your Sigma API client credentials.
export SIGMA_CLIENT_ID='<client-id>'
export SIGMA_CLIENT_SECRET='<client-secret>'
# The connection that points at the same data platform as your source content. Optional.
export SIGMA_CONNECTION_ID='<connection-id>'

For guidance on identifying your API base URL and generating client credentials, see Get started with the Sigma REST API. The skills also include a setup script that prompts for these values and stores them for you.

Run a migration

If you’re migrating multiple dashboards, start with the assessment skill for your source tool. The assessment is read-only, and produces a ranked list of priorities, so you can migrate the highest-value content first. If you already know which dashboards you want to migrate first, you can skip the assessment.

To start a migration, describe what you want in your own words. The more context you provide, the less the assistant has to guess. Include the source content, the destination folder, the connection to use, and how you want the assistant to handle formulas and security:

Migrate my Power BI report to Sigma.
Source: \[report URL or file path]
Land it in the Sigma folder “\[folder name or ID]”.
Use my Sigma connection “\[connection name or ID]”. It points at the same data platform the report reads from.

Before you build, tell me whether you’re reusing an existing Sigma data model or creating a new one. If the source data isn’t in the data platform yet, stop and tell me.

As you build, translate the real measures from the source. If you can’t convert a formula exactly, flag it instead of substituting something simpler.

Tell me what row-level and column-level security exists on the source, and wait for my decision before you apply anything.

When you’re done, show me the comparison between Sigma and the source, and list anything that didn’t carry over.

For example, the assistant asks whether to reuse an existing data model or create a new one, and whether to apply the security it found on the source content.

Review migrated content

The migration skills verify that certain summary values match between the source tool and the generated Sigma document. However this verification method cannot capture all potential considerations, and AI assistants can make mistakes. Before sharing generated documents with your organization, consider reviewing other details like the following:

Compare the full Sigma page against the source dashboard. Check the layout, chart types, sort order, and filters
Ask the assistant to list anything it couldn’t convert.
If you applied row-level or column-level security, check the document as a restricted user as well as an administrator. For more information, see Impersonate users.

If you choose not to apply the row-level or column-level security detected on your source content, everyone with access to the migrated workbook can see every row.