--- title: "Metabase Integration" description: "Run saved questions and ad-hoc queries, manage cards, collections, dashboards, and databases, and export results as JSON or CSV from your flows over the Metabase API." url: https://flowrunner.ai/integrations/metabase date_modified: 2026-08-01T02:40:32-07:00 --- # Metabase [Developer Tools](https://flowrunner.ai/integrations/category/developer-infrastructure) Run saved questions and ad-hoc queries, manage cards, collections, dashboards, and databases, and export results as JSON or CSV from your flows over the Metabase API. [Verified](https://flowrunner.ai/integrations/verified "What does verified mean?") · 22 actions · API key · available [Metabase website](https://metabase.com/) · [Platform Documentation](https://www.metabase.com/docs/latest/api-documentation) · Capability data verified 2026-08-12 1. A scheduled flow runs on the reporting cadence 2. Agent resolves the saved question (card) to run 3. Agent calls Run Card Query Export to pull the report as CSV 4. Agent confirms the export returned rows before handing it off 5. Agent emails the export to stakeholders or posts a summary to the channel 6. An analyst confirms before the agent deletes any shared card during cleanup ## What This Integration Enables FlowRunner agents use Metabase to move analytics into the rest of the operation without a person exporting by hand. An agent can run a saved question or an ad-hoc SQL query on a schedule and route the rows into another system, export results as JSON or CSV for reporting, email, or spreadsheet steps, and create and update cards, dashboards, and collections to keep BI content in sync. It can discover a database's tables and fields, then build a native SQL or structured MBQL query against it, and trigger a schema sync after upstream data-model changes. What it does not do on its own is delete a card, because a saved report others depend on cannot be restored. That confirmation stays with a person. ### Without FlowRunner **Manual report pulls**: An analyst opens Metabase and exports the same report every week **Copy-paste handoff**: Query results are pasted into email or a spreadsheet by hand **Risky cleanup**: Someone deletes an old card that a dashboard still depended on ### With FlowRunner **Scheduled exports**: The agent runs the saved question and exports the rows on cadence **Routed automatically**: The export flows straight into email, a sheet, or an alert **Confirmed deletes**: Deleting a shared card waits for an analyst to confirm ## Use Case Scenarios ### Scheduled report export On the reporting cadence, the agent calls Run Card Query Export to pull a saved report as CSV, confirms it returned rows, and emails the export to stakeholders. For a live tracking need, it instead calls Run Query for an ad-hoc SQL result and appends each row into a spreadsheet. The report lands in the right hands without an analyst running the export by hand. ### Anomaly alerting The agent calls Run Card Query on a key metric and compares the result against a threshold. When the number is anomalous, it posts a summary to the analytics channel with the figure and the card link. A metric that moves the wrong way surfaces the same day, not at the next review. ### BI content kept in sync with a deletion gate As data models change, the agent calls Sync Database Schema so Metabase reflects new tables, and it creates or updates cards and dashboards to match. During cleanup it flags cards that no dashboard references. It does not delete them. It lists them for an analyst and asks which to remove, and only deletes on confirmation. ## Human-in-Loop Highlight The human-in-loop moment in Metabase is the card deletion gate. Running queries, exporting results, syncing schemas, and creating or updating cards are all safe or reversible, so the agent runs them by rule. Deleting a card is not, because a saved question that dashboards and people rely on does not come back. When a cleanup flow flags cards with no references, the agent lists them and asks in the analytics channel: "These three cards are not used by any dashboard. Confirm which to delete?" An analyst confirms. Only then does the agent call Delete Card. The agent keeps BI content in sync on its own. A person owns the deletes that remove a shared report. Agent processes routinely Detects exception requiring judgment Clear match Continues automatically Ambiguous Routes to human via preferred channel Human decides Agent resumes with decision ## Agent Capabilities 22 actions ### Cards 7 - **Create Card** Create a saved question (card). - **Delete Card** Delete a card. Removing a shared report is not reversible, so run it behind a human confirmation. - **Get Card** Retrieve a card's definition. - **List Cards** List the saved questions in the instance. - **Run Card Query** Run a saved question and return its rows. Used to pull a scheduled report. - **Run Card Query Export** Run a saved question and export the results as JSON or CSV. - **Update Card** Update a card's query, name, or collection. ### Datasets 2 - **Export Query** Run an ad-hoc query and export the results as JSON or CSV. - **Run Query** Run an ad-hoc native SQL or structured MBQL query against a database. ### Databases 4 - **Get Database** Retrieve a database connection's details. - **Get Database Metadata** Retrieve a database's tables and fields. Used to discover a schema before building a query. - **List Databases** List the databases connected to the instance. - **Sync Database Schema** Trigger a schema sync so Metabase reflects new tables and columns. ### Collections 3 - **Create Collection** Create a collection to organize cards and dashboards. - **Get Collection Items** List the items inside a collection, such as cards, dashboards, and sub-collections. - **List Collections** List the collections in the instance. ### Dashboards 3 - **Create Dashboard** Create a dashboard. - **Get Dashboard** Retrieve a dashboard's definition. - **List Dashboards** List the dashboards in the instance. ### Users & Health 3 - **Get Current User** Retrieve the authenticated user. Used as a connection check. - **Health Check** Check the instance health. - **List Users** List the users in the instance. Requires an admin-level key. ## Frequently Asked Questions ### What can FlowRunner do with Metabase? FlowRunner agents can run List Cards, Get Card, and Create Card in Metabase, plus 19 more actions. ### Does connecting Metabase to FlowRunner require OAuth? No. Metabase connects to FlowRunner with an API key, no OAuth flow required. ### Can Metabase trigger a FlowRunner workflow automatically? Metabase doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger. **Work at Metabase?** This integration exposes Metabase to AI agents on every FlowRunner plan, including through MCP, at no cost to you. [See what FlowRunner offers integration partners](https://flowrunner.ai/integrations/partners), including how to keep this page current. --- Markdown version of https://flowrunner.ai/integrations/metabase. Site index: https://flowrunner.ai/llms.txt