Model Context Protocol
Give Claude, ChatGPT, Cursor, or any MCP client the power to create real PostgreSQL databases and run SQL — through one remote MCP server. No Docker, no local install, no infrastructure to run.
https://api.dbaas.dev/v1/mcpAuthentication via API key — sign in free and create one under Settings → API Keys.
"Create a postgres database for this project" — a real, managed PostgreSQL instance spins up in seconds and the connection string comes back to the model.
Schema design, data loading, queries, migrations — the model executes SQL against its databases and reads results without leaving the conversation.
List databases, explore schemas, check status, trigger backups — full lifecycle management exposed as MCP tools.
# Add the DBaaS Postgres MCP server to Claude Code
claude mcp add --transport http dbaas https://api.dbaas.dev/v1/mcp \
--header "Authorization: Bearer YOUR_API_KEY"
# Then just ask:
# "Create a postgres database and load this CSV into it"
# "Show me the schema of my orders database"
# "Run: SELECT count(*) FROM users WHERE created_at > now() - interval '7 days'"// .cursor/mcp.json — Cursor / Windsurf / any JSON-config MCP client
{
"mcpServers": {
"dbaas-postgres": {
"url": "https://api.dbaas.dev/v1/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}1. Open Claude.ai → Settings → Connectors → "Add custom connector"
2. Name: DBaaS Postgres
3. URL: https://api.dbaas.dev/v1/mcp
4. Paste your API key when prompted
5. Ask Claude: "Spin up a postgres database and create a users table"Full tool reference and troubleshooting in the MCP integration docs.
Most Postgres MCP servers are local processes: install Node or Docker, manage credentials in config files, keep a database running on your machine. That works for one developer on one laptop — it breaks down the moment your agent runs in the cloud, your team shares a workflow, or you want Claude.ai (which only speaks to remote servers) involved.
DBaaS.dev's MCP server is remote-first: one URL, API-key auth, and the databases it creates are real managed PostgreSQL instances with backups, TLS, and a dashboard — not throwaway containers on your laptop. Your agent can create a database in a conversation today and your application can still be using it in production next month.
Need disposable databases instead? The same platform offers ephemeral Postgres with automatic TTL cleanup — ideal for tests and one-shot agent tasks. And if you're building agent systems end-to-end, see the AI agent database overview for REST and A2A options alongside MCP.
Yes — the free tier includes managed PostgreSQL databases and full MCP access. You need a free account to create an API key; no credit card required.
Any client that supports remote (HTTP) MCP servers: Claude.ai custom connectors, ChatGPT, Claude Code, Cursor, Windsurf, and the MCP SDKs. Local-only clients can bridge via mcp-remote.
Database provisioning, SQL execution, schema inspection, database listing and lifecycle management — everything an agent needs to create and use PostgreSQL autonomously.
No. It's a hosted remote MCP server — one URL plus your API key. No Docker, no npm install, no local Postgres.
Yes — managed PostgreSQL 17 instances with TLS, automatic backups, and a web dashboard. Connection strings work with any Postgres driver, so what your agent builds, your app can keep using.
Yes — ephemeral databases with a TTL (auto-deleted after expiry) are available via the REST API and are perfect for tests and one-shot jobs.