October 14, 2026
API, MCP and code updates
Entities in the API and MCP
Now you can access row-level data from the Steep API and MCP, so your own tools and agents can work with the individual records behind your metrics, not only the aggregates. Entity definitions work in the MCP right away. Reading entity data requires reauthenticating your MCP connection.
Multiple breakdowns in the API
Query a metric with any number of breakdowns in a single API request, so your custom apps can build visualizations that combine several dimensions at once, without stitching together separate calls.
Sync and preview code changes from the API
Trigger a sync of your semantic models from the API, whether they're defined in Steep as code, dbt, or Cube. The new POST /v1/code/syncs endpoint queues the sync, and you can check its status until it completes or fails.
Add dryRun: true to preview which definitions a sync would create, update or remove, along with any errors or warnings, without changing anything. For Steep as code, include your YAML files in the request to check local changes before pushing them to GitHub.
Instant feedback when working in code
Your Steep YAML files now validate as you type. Add this line to the top of any file, and editors with YAML language server support, like VS Code and Cursor, flag errors and suggest fields in real time:
# yaml-language-server: $schema=https://api.steep.app/v1/code/schema
For stricter checks, point module files to /schema/module and targets files to /schema/targets. See the help center for details.
Improvements & fixes
APIThe API can now create reports with the new tablesAPIUnrecognized block settings now return an error instead of being ignoredOAuthClearer errors for invalid clients and missing OpenID configuration, and up to 100 redirect URIs per client