Metabase
Metabase is an open-source business intelligence tool that lets you ask questions about your data and visualize answers as charts, graphs, and dashboards.
Verdict
Common use cases
- Duplicate client dashboards with custom names
- Bookmark key reports for team quick access
- Clear cached queries after data refreshes
- Organize collections and move dashboards
- Delete outdated actions and clean workspace
Integration
- Vendor
- Metabase
- Category
- other
- Auth
- API_KEY
- Tools
- 50
- Composio slug
metabase
Tools
- Copy Dashboard
Tool to create a copy of an existing Metabase dashboard. Use when you need to duplicate a dashboard with optional customization of name, description, and collection placement. Supports both shallow copy (references same cards) and deep copy
- Create Bookmark
Tool to create a new bookmark for a Metabase item (card, dashboard, collection, or dataset). Use when the user wants to bookmark a specific item for quick access.
- Create Bookmark
Tool to create a new bookmark for a Metabase item (card, dashboard, collection, or dataset). Use when you need to bookmark a specific item for the authenticated user.
- Create Field Dimension
Tool to set the dimension for a field at the specified ID in Metabase. Use when you need to configure field remapping or display values for better data presentation.
- Delete Actiondestructive
Tool to delete an Action from Metabase. Use when you need to permanently remove an action. This operation cannot be undone.
- Delete Actiondestructive
Tool to delete a Metabase action by its ID. Use when you need to permanently remove an action from the system.
- Delete API Cachedestructive
Tool to delete/clear cached data in Metabase for specific entities. Use when you need to invalidate cache for a database, dashboard, question, or clear all cache (root).
- Delete Bookmarkdestructive
Tool to delete a bookmark from a Metabase item (card, dashboard, or collection). Use when you need to remove a bookmark from a specific item.
- Delete Bookmarkdestructive
Tool to delete a bookmark by model and ID. Use when you need to remove a bookmark assigned to the user making the request. The operation returns HTTP 204 No Content on successful deletion.
- Delete Carddestructive
Tool to delete a Metabase card (question) by its ID. Use when you need to permanently remove a card. Note: This endpoint is deprecated - consider archiving cards instead.
- Delete Carddestructive
Tool to hard delete a Metabase card. Use when you need to permanently remove a card from the system. For soft deletion, use PUT /api/card/:id instead.
- Delete Dashboarddestructive
Tool to delete a Metabase dashboard by its ID. Use when you need to permanently remove a dashboard.
- Delete Dashboarddestructive
Tool to hard delete a Metabase dashboard by its ID. Use when you need to permanently remove a dashboard. This will also remove any questions/models/segments/metrics that use this database. For soft delete, use the update dashboard action in
- Delete Field Dimensiondestructive
Tool to remove the dimension associated with a field at the specified ID. Use when you need to disassociate a dimension from a field in Metabase.
- Delete Glossary Entrydestructive
Tool to delete a glossary entry from Metabase. Use when you need to permanently remove a glossary entry by its ID.
- Delete Glossary Itemdestructive
Tool to delete a glossary item by ID. Use when you need to remove a glossary term from Metabase. The operation returns HTTP 204 No Content on successful deletion.
- Delete Model Indexdestructive
Tool to delete a Metabase model index by its ID. Use when you need to permanently remove a model index entry.
- Delete Model Indexdestructive
Tool to delete a ModelIndex by ID from Metabase. Use when you need to remove a model index from the system.
- Delete Timelinedestructive
Tool to delete a Metabase timeline and cascade delete all its associated events. Use when you need to permanently remove a timeline. Deletion is permanent and cannot be undone.
- Delete Timelinedestructive
Tool to delete a timeline by ID. Use when you need to permanently remove a timeline. Returns success status on completion.
- Delete Timeline Eventdestructive
Tool to delete a timeline event by ID. Use when you need to permanently remove a specific timeline event from Metabase.
- Delete Timeline Eventdestructive
Tool to delete a Metabase timeline event. Use when you need to permanently remove a timeline event from the system.
- Delete User Key-Valuedestructive
Tool to delete a user key-value pair from Metabase. Use when you need to remove a stored key-value entry within a specific namespace.
- Delete User Key-Value Pairdestructive
Tool to delete a user key-value pair by namespace and key. Use when you need to remove a specific key-value entry from a user's namespace.
- Delete User Subscriptionsdestructive
Tool to delete all Alert and DashboardSubscription subscriptions for a User. Use when you need to remove all subscriptions for a specific user and archive alerts/dashboards they created. This is an enterprise feature requiring admin privile
- Delete User Subscriptionsdestructive
Tool to delete all Alert and DashboardSubscription subscriptions for a User. Also archives all Alerts and DashboardSubscriptions created by the User. Requires premium token with :audit-app feature. Only allowed for admins or current user.
- Disable Database Model Persistence
Tool to disable model persistence for a Metabase database. Use when you need to turn off caching of model results. Returns success whether persistence was already disabled or not (idempotent operation).
- Discard Database Field Values
Tool to discard all saved field values for a specific database. Use when you need to clear cached field value data for all fields in a database. Requires superuser privileges.
- Discard Field Values
Tool to discard the FieldValues belonging to a Metabase field. Use when you need to clear cached field values. Only applies to fields that have FieldValues. If the Field's Database is set up to automatically sync FieldValues, they will be r
- Discard Table Field Values
Tool to discard the FieldValues belonging to the Fields in a Metabase Table. Use when you need to clear cached field values for all fields in a table. Only applies to fields that have FieldValues. If this Table's Database is set up to autom
- Dismiss Database Sync Spinner
Tool to manually set the initial sync status of a Database and corresponding tables to complete. Use when you need to dismiss the synchronization spinner for a database after initial setup.
- Get Action
Tool to retrieve a Metabase action by ID. Use when you need to fetch details about a specific action including its configuration, parameters, and metadata.
- Get Action Execute Parameters
Tool to fetch values for filling in action execution parameters. Use when you need to get parameter values for executing an action by providing PK parameters and values to select.
- Get All Alerts
Tool to retrieve all alerts in Metabase. Use when you need to get a list of all configured alerts with their settings and notification channels.
- Get all Metabase actions
Tool to retrieve all actions available in the Metabase instance. Actions are entities that let you build custom forms and business logic with parameterized SQL that writes back to your database. Returns an array of action objects including
- Get Automagic Dashboard Candidates
Tool to retrieve automagic dashboard candidates for a specific database. Use when you need to discover interesting tables and schemas for automatic dashboard generation. Returns an array of schema candidates with their tables, ordered by in
- Get Automagic Dashboard Cell Analysis
Tool to return an automagic dashboard analyzing a specific cell in an entity's data. Use when you need to generate automated insights and visualizations for a cell query within a table, database, or other Metabase entity. The cell-query mus
- Get Popular Items
Tool to get the list of 5 popular items on the Metabase instance. Use when you need to discover frequently accessed dashboards or cards. Returns both dashboards and cards with their metadata including name, description, parent collection, a
- Get Recent Activity Views
Tool to retrieve recently viewed items in Metabase. Use when you need to get a list of cards, dashboards, collections, tables, or documents that the authenticated user has recently accessed. Returns metadata including timestamps and permiss
- Rescan Table Field Values
Tool to manually trigger an update for FieldValues for Fields belonging to a specific Table. Use when you need to refresh field value caches after table data changes. Only applies to Fields that are eligible for FieldValues.
- Run Card Pivot Query
Tool to run a pivot table query on a Metabase card (question). Use when you need to execute the query associated with a specific card and get results formatted for pivot table visualization. The response includes data rows, column metadata,
- Run Card Query
Tool to run the query associated with a Card and return its results. Use when you need to execute a card's query with optional parameters and cache control.
- Run Dashboard Pivot DashCard Query
Tool to run a pivot table query for a specific DashCard. Use when you need to execute a pivot table query on a dashboard card with optional parameter filters.
- Save Dashboard to Collection
Tool to save a denormalized dashboard description into a collection. Use when you need to create a new dashboard directly in a specific collection with full configuration including cards, tabs, and parameters.
- Search Dataset Parameter Values
Tool to search for parameter values in datasets when editing cards or dashboards. Use when you need to provide autocomplete functionality for parameters. Expects a query string and returns matching values.
- Sync Table Schema
Tool to trigger a manual update of the schema metadata for a specific Table. Use when you need to refresh table schema information after database changes. This synchronizes columns, data types, and other metadata from the underlying databas
- Trigger Database Field Values Rescan
Tool to trigger a manual scan of field values for a Database. Use when you need to refresh field values for filter dropdowns after data changes. Requires superuser permissions.
- Trigger Database Schema Sync
Tool to trigger a manual update of the schema metadata for a Database. Use when you need to refresh the database schema metadata in Metabase after structural changes.
- Unpersist Card
Tool to unpersist a Metabase card/model. Deletes the persisted table backing the model and all queries after this will use the card's query rather than the saved version. Use when you need to remove persistence from a model.
- Update Field Values
Tool to update field values and human-readable values for a Field whose semantic type is category/city/state/country or whose base type is type/Boolean. Use when you need to set or update the distinct values for a field along with optional
Setup
Setup guide
- 11. In Switchy, open your workspace settings and navigate to the Integrations tab. 2. Find Metabase in the MCP directory and click Connect. 3. You'll be prompted to enter your Metabase instance URL (e.g., https://analytics.yourcompany.com) and an API key. 4. Generate the API key in Metabase by going to Admin Settings → API Keys → Create Key, then copy the token. 5. Paste the URL and key into Switchy's connection form and click Authorize. 6. Switchy will verify the connection by fetching your Metabase user profile. 7. Once connected, open any Project and type '@Metabase' followed by a task like 'copy the Q4 Sales dashboard' to test the integration. 8. The AI will confirm available tools and execute your request, returning results directly in the chat.
What teammates see: by default, memories from Metabase are scoped to the Project (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Projects have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Copy Dashboard for Client
@Metabase copy dashboard ID 42 and name it 'Q1 2025 Client Review Dashboard' in the Client Reports collection
Bookmark Team Reports
@Metabase bookmark dashboard ID 18 and card ID 203 for quick access to our weekly metrics
Clear Stale Cache
@Metabase delete the cache for database ID 5 since we just refreshed the warehouse tables
Organize Dashboard Collections
@Metabase move dashboard ID 31 to the Marketing collection and update its description to reflect Q4 campaigns
Remove Obsolete Actions
@Metabase delete action ID 12 because that automation was replaced by our new pipeline
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@metabase show me all dashboards in the 'Sales Analytics' collection and which ones are bookmarked by our team
Found 7 dashboards in the Sales Analytics collection. Three are currently bookmarked: 'Q4 Revenue Breakdown' (bookmarked by 4 users), 'Regional Performance' (3 users), and 'Pipeline Velocity' (2 users). The remaining dashboards include 'Customer Segmentation', 'Deal Funnel', 'Rep Leaderboard', and 'Forecast vs Actuals'. All dashboards were last modified within the past 30 days, with 'Pipeline Velocity' seeing the most recent update 3 days ago.
This example demonstrates read-only collection browsing and bookmark querying. Requires API key with read access to the target collection. Response reflects account state at query time—dashboard counts and bookmark status will vary by organization. No side effects.
@metabase copy the 'Weekly Metrics' dashboard to the 'Executive Reports' collection and rename it 'Board Meeting Metrics'
Successfully created a copy of 'Weekly Metrics' dashboard. The new dashboard 'Board Meeting Metrics' has been placed in the Executive Reports collection with ID 42. All 8 cards from the original dashboard have been duplicated, maintaining their layout and visualization settings. The copy is now accessible at /dashboard/156. Note that any SQL queries or data model references remain pointed to the same underlying sources as the original.
This write operation creates a new dashboard resource in Metabase. Requires API key with write permissions to the destination collection. The copy is shallow—cards are duplicated but share the same data sources and queries. Users should verify collection permissions after copying to ensure intended access control.
@metabase analyze the last 10 questions created in our Marketing collection and suggest which ones should be combined into a new dashboard
Reviewed 10 recent questions in Marketing collection. Recommend grouping 5 questions into a 'Campaign Performance' dashboard: 'Email Open Rates by Segment' and 'Click-Through Trends' show complementary engagement metrics; 'Ad Spend vs Conversions' and 'Cost Per Acquisition by Channel' provide budget efficiency views; 'Landing Page Performance' bridges both themes. These questions share date range filters and would benefit from unified time controls. The remaining 5 questions cover disparate topics (SEO rankings, social media sentiment, content calendar, brand mentions, competitor tracking) and are better suited for separate dashboards or ad-hoc analysis.
This synthesis example combines Metabase's question metadata retrieval with AI reasoning to identify dashboard opportunities. The MCP fetches question titles, descriptions, and filter configurations—the AI interprets thematic relationships. Requires read access to the collection. Recommendations are advisory; actual dashboard creation would require a follow-up action.
Use-case deep-dives
When Metabase MCP beats manual dashboard cloning for A/B tests
A 6-person product team runs weekly experiments on their metrics dashboards—testing new KPI layouts, swapping chart types, or filtering by cohort. Without this MCP, they manually duplicate dashboards in the Metabase UI, which takes 4-5 clicks per copy and loses track of which version tested what. The Copy Dashboard tool lets an AI agent clone a dashboard, rename it with the experiment tag, and drop it in a dated collection—all in one prompt during standup. The 50-tool scope also covers bookmark management, so the team lead can auto-bookmark the winning version for stakeholders. This works until you hit 20+ dashboards per sprint; beyond that, you need a proper versioning system outside Metabase. If your team runs 2-4 dashboard experiments a week and wants to skip the UI tedium, this MCP pays off immediately.
Why Metabase MCP matters for support teams syncing stale data
A 3-person support team uses Metabase dashboards to answer customer questions about usage trends and account health. The dashboards pull from a nightly ETL, but when a customer asks about today's data, the team has to manually clear the cache in Metabase settings—a 30-second detour that breaks their flow. The Delete API Cache tool lets an AI agent invalidate the cache for a specific dashboard or question on demand, so the support rep just says "refresh the churn dashboard" and gets live data in the next query. The API key auth means no OAuth dance for each agent session. This setup degrades if your Metabase instance has 100+ concurrent users hammering the cache; at that scale, you want a dedicated cache-warming job. For small support teams fielding 10-20 data questions a day, this MCP turns cache management into a one-liner.
When this MCP wins for post-sprint dashboard housekeeping
A solo data analyst at a 12-person startup builds 8-10 exploratory dashboards per sprint, testing hypotheses for the exec team. By sprint-end, half are obsolete, but deleting them in the Metabase UI means clicking through each dashboard's settings menu. The Delete Action tool (plus the broader 50-tool set) lets the analyst tell an AI agent "archive all dashboards tagged 'experiment' from last sprint" and bulk-remove the clutter in one go. The MCP also handles field dimension updates, so the analyst can tweak how columns display across dashboards without opening the UI. This breaks down if your Metabase has strict RBAC and the analyst doesn't own all the dashboards—API key auth won't override permissions. If you're a one-person analytics team drowning in throwaway dashboards, this MCP cuts cleanup time from 20 minutes to 2.
Frequently asked
What can the Metabase MCP do in Switchy?
It connects your team's Metabase instance so AI agents can manage dashboards, bookmarks, and cached data without opening the Metabase UI. Agents can copy dashboards, create bookmarks for frequently-used reports, configure field dimensions for better data presentation, delete actions, and clear API caches. You still run queries and build visualisations in Metabase itself — the MCP handles the metadata and organisational tasks.
Do I need admin access to connect Metabase MCP?
You need an API key from your Metabase instance, which typically requires admin or elevated permissions to generate. The key's scope determines what the MCP can do — if your key only has read access, agents won't be able to delete actions or copy dashboards. Check with your Metabase admin about key permissions before connecting.
Can the Metabase MCP run SQL queries or create new charts?
No. The MCP manages existing Metabase objects — copying dashboards, bookmarking cards, clearing caches — but it doesn't execute queries or build visualisations. If you need agents to query your database directly, connect a database MCP (Postgres, MySQL) instead. Use Metabase MCP when you want agents to organise or duplicate work that already exists in Metabase.
Why use this instead of just opening Metabase?
It's faster for repetitive admin tasks. If your team regularly copies dashboards for new clients, bookmarks reports for onboarding, or clears caches after data refreshes, agents can do it in Switchy without context-switching. You're not replacing Metabase — you're automating the busywork around it. For one-off tasks, opening Metabase is still simpler.
Who on the team should connect the Metabase MCP?
Whoever manages your Metabase instance and can generate API keys. That's usually a data analyst, BI lead, or engineering lead. They'll need to decide which permissions the key should have — read-only if you only want agents to bookmark things, full access if you want them to delete actions or copy dashboards. The connection lives in your Switchy workspace, so everyone can use it once it's set up.