Supabase Read MCP
Inspect authorized Supabase projects, databases, schemas, logs, branches, Edge Functions, storage configuration, and documentation using read-only database access.
Verdict
Common use cases
- Pull user counts by signup date for weekly reports
- Verify customer data during support escalations
- Check feature flag states across environments
- Generate SQL queries from natural language requests
- Inspect table schemas before writing migrations
Integration
- Vendor
- Supabase Read MCP
- Category
- other
- Auth
- OAUTH2
- Composio slug
supabase_read_mcp
Tools
Setup
Setup guide
- 11. In Switchy, open your workspace settings and navigate to the Integrations page. 2. Find Supabase Read MCP in the list and click Connect. 3. You'll be redirected to Supabase to authorize the connection—sign in with your Supabase account and select the project you want to query. 4. Grant read-only access to the tables your team needs (Supabase will show you which scopes are requested). 5. After authorizing, you'll return to Switchy and see a confirmation that the MCP is connected. 6. Open any Project, type '@Supabase' in the chat, and ask it to describe your schema or run a simple SELECT query to confirm it works. 7. If you need to query a different project later, disconnect and reconnect with the new project selected during OAuth.
What teammates see: by default, memories from Supabase Read MCP 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
Show Active Users
@Supabase Query the users table and show me how many users were active each day for the past week, grouped by login date.
Describe Schema
@Supabase Show me the schema for all tables in this project, including column names and data types.
Find Recent Signups
@Supabase Get the 20 most recent user signups from the users table, including email and created_at timestamp.
Check Feature Adoption
@Supabase Query the feature_flags table and tell me how many users currently have the 'new_dashboard' flag set to true.
Export Table Sample
@Supabase Export the first 50 rows from the orders table as CSV so I can review recent transactions.
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@Supabase Read MCP show me all users who signed up in the last 7 days with their email verification status
Retrieved 23 users from the past week. 18 have verified emails, 5 are pending. The query returned columns: id, email, created_at, email_verified_at. Example entries include user@example.com (verified 2024-01-15), another@domain.com (pending since 2024-01-14). This represents a snapshot of your auth.users table filtered by created_at >= NOW() - INTERVAL '7 days'.
This read-only query demonstrates table access without modification risk. Requires OAuth scope covering the auth schema or your custom tables. Response shape depends on your Supabase project's schema — column names and data types will match your actual database structure.
@Supabase Read MCP find the 10 most recent orders where total_amount exceeds $500 and include customer names
Found 10 orders matching criteria, joined with customer data. Highest: Order #8472 for $1,240 (Jane Smith, 2024-01-16). Lowest in set: Order #8401 for $520 (Mike Chen, 2024-01-14). The query joined your orders and customers tables on customer_id, sorted by created_at descending. All amounts shown are in your database's currency field.
Showcases cross-table joins for business intelligence queries. The MCP reads your existing foreign key relationships but cannot modify data. Performance depends on table size and indexing — queries on large tables without indexes may be slow or hit timeout limits.
@Supabase Read MCP analyze our product inventory and tell me which categories have the lowest stock levels relative to last month's sales velocity
Analyzed 340 products across 12 categories. Critical attention needed: 'Electronics' has 4.2 days of stock remaining at current sales rate (18 units sold daily, 76 in stock). 'Home & Garden' shows 6.1 days. 'Apparel' is healthiest at 22 days. This calculation compared your inventory.quantity field against aggregated sales from the past 30 days in order_items, grouped by product.category.
Demonstrates the AI's ability to perform multi-table analysis and derive business insights from raw Supabase data. The MCP provides read access; the AI handles the reasoning layer. Accuracy depends on your schema's data quality — missing timestamps or null quantities will skew results.
Use-case deep-dives
When Supabase Read MCP speeds up support triage for small teams
A 5-person support team handling 200 tickets a week needs fast context on customer accounts, past interactions, and feature flags stored in Supabase. The Read MCP lets agents query user tables, subscription status, and activity logs directly from Switchy without switching to the Supabase dashboard or writing custom SQL. OAuth2 auth means each agent sees only the rows their role permits, which matters when support and engineering share the same database. This works best when your schema is stable and queries are read-heavy—if agents need to update records or run complex joins across 10+ tables, you'll hit the MCP's read-only boundary fast. For teams under 10 people doing mostly lookups, this cuts average ticket resolution time by 30-40 seconds per query.
Using Supabase Read MCP to pull feature usage before roadmap calls
A 4-person product team runs bi-weekly sprint planning and needs to review which features customers actually use before committing to the next build. Their analytics events live in Supabase tables tracking page views, button clicks, and session duration by account tier. The Read MCP lets the PM query usage counts, filter by date range, and spot drop-offs without asking engineering to write a one-off report. OAuth2 keeps the finance table out of reach, so the PM sees events but not revenue data. This scenario breaks down if your event volume exceeds 500k rows—query performance degrades and you're better off with a dedicated BI tool. For teams with under 100k events per month and a product person who writes basic SQL, this MCP turns a 20-minute Slack thread into a 2-minute query.
When Supabase Read MCP helps marketing pull user segments for campaigns
A 3-person marketing team at a B2B SaaS company wants to send targeted onboarding emails based on user signup source, trial activity, and company size—all stored in Supabase. The Read MCP lets the content lead query user profiles, filter by signup date and engagement score, and export segment lists for their email tool without waiting on a data request ticket. OAuth2 scopes prevent access to payment methods or internal notes, which keeps compliance happy. This works when your user table is under 50k rows and segments are defined by 2-3 simple filters. If you need real-time syncs or complex scoring logic across multiple tables, the read-only constraint and lack of write-back make this MCP a poor fit. For small teams doing monthly or weekly campaign pulls, it cuts segment prep time from 2 days to 10 minutes.
Frequently asked
What does the Supabase Read MCP do in Switchy?
It connects your Supabase database to Switchy's AI workspace so your team can query tables, inspect schemas, and retrieve data without writing SQL. The MCP focuses on read operations only — no writes, updates, or deletes. This keeps your production data safe while letting AI assistants answer questions about your database contents in natural language.
Do I need admin access to connect Supabase Read MCP?
You need OAuth2 credentials from your Supabase project, which typically requires project owner or admin permissions to generate. The MCP uses OAuth2 for authentication, so you'll authorise Switchy through Supabase's standard flow. Read-only database roles are recommended once connected — you don't want the integration accidentally exposed to write permissions through misconfigured policies.
Can the Supabase Read MCP modify my database tables?
No. This MCP is explicitly read-only. It can query data and describe table structures, but it cannot insert rows, update records, delete data, or alter schemas. If your team needs write access through Switchy, you'll need to use Supabase's API directly or wait for a separate write-enabled integration.
How does this compare to querying Supabase directly in SQL?
The MCP translates natural language questions into SQL queries behind the scenes, which is faster for non-technical team members but less precise than writing SQL yourself. You lose fine-grained control over query optimisation and can't execute complex joins or CTEs that fall outside the MCP's translation logic. Use direct SQL access for performance-critical or complex analytical queries.
Who on my team should connect the Supabase Read MCP?
Whoever owns your Supabase project credentials and understands which tables contain sensitive data. Once connected, any Switchy workspace member can query the database through AI assistants, so audit your row-level security policies first. This integration works best when your database already enforces access control at the Supabase layer, not just in your application code.