Software decision profile
Supabase MCP
Exposes PostgreSQL database tables and pgvector search natively to agents.
Explore the trade-offs.
Where it stands outDirect DB accessExplore strengths ↗Close details ↑
- pgvector support
What to weighPrompt injection is a documented risk: untrusted content in the database can steer the agentExplore limitations ↗Close details ↑
- Supabase advises connecting to a production project only when the task requires production evidence, using project scoping, read-only mode and restricted feature groups
Best suited toDevelopers using AI coding assistants to build and migrate Supabase projects
What Supabase MCP’s own pages state.
Each line was read on the vendor’s site and found word for word in the quoted page. It records what the vendor states, not a test by Fathom Layer.
- Self-hosted
- Yes
Where it wins, where it doesn't
Pros
- Direct DB access
- pgvector support
Cons
- Prompt injection is a documented risk: untrusted content in the database can steer the agent
- Supabase advises connecting to a production project only when the task requires production evidence, using project scoping, read-only mode and restricted feature groups
Editorial note
The Supabase MCP server lets AI clients such as Claude, Cursor and Windsurf create projects, design tables, run SQL, apply migrations and read logs and advisors. Supabase's own docs flag prompt injection as the main risk and advise connecting to a production project only when the task requires production evidence, using project scoping, read-only mode and restricted feature groups. That makes it a development tool first: useful for building, risky if pointed at live data carelessly.
Capabilities, limitations and fit.
In-Depth Review
The Supabase MCP server exposes Postgres tables and pgvector similarity search directly to an agent, so a model can query your application database — and your embeddings — as a native capability rather than through a hand-built tool.
What it unlocks
- Structured lookups. "How many active accounts signed up last week" resolves against the real schema.
- Vector search. With pgvector exposed, an agent can do semantic retrieval over your own content without a separate vector store.
That combination makes it a compact backbone for retrieval-augmented workflows on data you already hold in Supabase.
The caveat
Direct database access from a model is powerful and needs scoping. Use a role with only the permissions the task requires, keep row-level security enforced, and gate any write path behind human review.
Who should use it
Teams already on Supabase who want agents to read application data and do semantic search over their own content. If your data lives elsewhere, a database-specific connector is the better fit.
Frequently Asked Questions
Who is Supabase MCP for?↓
What are the drawbacks of Supabase MCP?↓
What does Supabase MCP do well?↓
How much does Supabase MCP cost?↓
Alternatives to consider
See all alternatives →Further reading
Featured badge
Building this product? Add the badge to your site to show it’s in the index.
<a href="https://fathomlayer.com/intelligence/mcp-servers/supabase-mcp" target="_blank" rel="noopener noreferrer"><img src="https://fathomlayer.com/fathom-badge.svg" alt="Featured on Fathom Layer" width="250" height="54" /></a>