Summarize a saved question’s data
Run saved Metabase question [question]. Summarize [business question] and mention any limits on the results reviewed.
metabase
Metabase AI Employee
Leina connects to Metabase to help you run saved reports, summarize business data, and inspect dashboards. Ask in your team chat. Leina does the work and brings back the result.
Connect Metabase, choose what to share, and hand off your first task.
Leina + Metabase Example Result
You
Run our saved “Orders by channel” question [question identifier]. Summarize which channels contributed the most orders.
Leina
I ran the saved question and summarized the channel mix from its results.
Metabase · Report summary
Orders by channel
A concise view of the saved report’s results.
- Direct
- 120 orders · Largest channel in this result
- Referral
- 80 orders · Second-largest channel
What Is a Metabase AI Agent?
A Metabase AI agent runs accessible saved questions and explains their results in business terms, keeping the reporting scope clear.
You specify the source and the destination. Leina uses the content available to your connection and returns the result for you to review.
Metabase
Metabase helps teams explore data through saved questions, charts, and dashboards.
-
Choose the report
Provide the saved question you want to review.
-
Run the question
Read its definition and retrieve the results.
-
Explain the findings
Summarize the figures within the available result set.
What to Ask Leina to Do with Metabase
Copy a prompt, add your details, and send it to Leina in your connected team chat.
Your next Metabase task can start with a message.
Connect. Ask. Get the Result.
Tell Leina what you need in your team chat. Include the source, destination, and expected result.
-
Connect Your Account
Connect the tools involved and give Leina access to the content it needs.
-
Try a Task
Share the source material, say where the result belongs, and describe what you need.
-
Check the Result
Review the result and any changes in the connected app.
Put Leina to Work in Metabase
Hand off these everyday tasks to your AI employee.
Get Current User
Get the Metabase user associated with the API key.
List Databases
List Metabase databases visible to the API key.
Get Database
Retrieve one Metabase database by ID.
List Collections
List Metabase collections visible to the API key.
Get Collection
Retrieve one Metabase collection by ID.
List Cards
List Metabase cards, also known as questions.
Get Card
Retrieve one Metabase card by ID.
List Dashboards
List Metabase dashboards visible to the API key.
Get Dashboard
Retrieve one Metabase dashboard by ID.
Search
Search Metabase content visible to the API key.
Search Content
Search native Metabase MCP content using keyword or semantic queries (not the REST search action). Requires native MCP on the instance.
Read Resource
Read up to five Metabase entity URIs. Preserves individual resource errors; list endpoints cap at 25 items.
Construct Query
Construct portable MBQL 5 without executing it; returns a stored query handle.
Construct Native Query
Construct SQL for saving as a question. This handle cannot be run by execute_query or query; use execute_sql to execute SQL.
Query
Execute MBQL or fetch its next page. Native MCP returns pages of 200 rows within a 2,000-row total budget.
Execute Query
Execute constructed MBQL and return rows and columns. Native SQL handles are not supported.
Execute Sql
Execute native SQL. Requires native-query permission and the instance's execute-SQL setting to be enabled.
Execute Question
Run a saved question. Parameterized questions and input template tags are not supported by native MCP.
Create Question
Save a constructed MBQL or SQL query as a question.
Update Question
Patch a saved question, optionally replacing its query with a stored handle.
Create Metric
Save constructed MBQL as a metric: one aggregation and at most one date/datetime grouping.
Update Metric
Patch a metric; replacement queries must still meet metric requirements.
Create Dashboard
Create a dashboard with optional saved questions, automatically positioned.
Update Dashboard
Patch a dashboard and apply ordered add/remove/move card mutations.
Create Collection
Create a collection, optionally nested under a parent.
List Mcp Tools
Discover native MCP tools and their upstream schemas. Unknown tools are discoverable, not automatically executable.
List Mcp Resources
Discover native MCP resources. Only documentation resources can be read by this connector; Apps are not supported.
Read Mcp Resource
Read native metabase://docs/ documentation, not entities or MCP Apps resources.
More Tools for Your AI Employee
Connect the Apps your work depends on, and give Leina the context to carry the task forward.
Keep the Conversation Moving. Let Leina Pick Up the Work.
Hand off a task while your team is discussing it. Leina uses your connected tools and brings the result back to the conversation.
See How Leina Works in SlackLeinaExample Result I ran the saved question and summarized the channel mix from its results.
Orders by channel
Metabase + Leina: Your Questions, Answered
Can I run a report that asks for filter values?
The supported saved-question execution does not accept questions requiring parameters. Choose a report with its criteria already defined.
Will a summary include every row?
A query can return up to 2,000 rows through this connection. Larger datasets need a more focused saved question.
What must be enabled on my instance?
Metabase’s native MCP features must be available and enabled for these query operations.
Make Leina Your Metabase AI Employee
Start with one task you already need to do. Tell Leina what a finished result should look like.
Connect Metabase, choose what to share, and hand off your first task.