Postgres FAQs

Question 1

What is Postgres?

Accepted Answer

Postgres (or postgres-mcp) is a production-grade server and Go library that provides AI agents and LLMs safe, controlled access to PostgreSQL databases through the Model Context Protocol (MCP).

Question 2

How does Postgres prevent SQL injection and ensure query safety for AI?

Accepted Answer

It implements protocol-level enforcement that only allows single statements, plus 23 AST-based protection rules using PostgreSQL's C parser to detect and block disallowed operations, ensuring AI queries are secure.

Question 3

What deployment options does Postgres offer?

Accepted Answer

Postgres can run as a standalone HTTP server, making it easy to integrate with various AI tools, or be embedded directly as a Go library within your application for custom implementations and tighter control.

Question 4

Can Postgres sanitize sensitive data before returning results to LLMs?

Accepted Answer

Yes, it includes robust regex-based field-level data sanitization, allowing you to mask or replace sensitive information (like PII) within JSON results before they are exposed to AI agents.

Question 5

Is Postgres safe to expose to the public internet?

Accepted Answer

No, Postgres is explicitly designed for local or internal network use only. It intentionally lacks authentication and CORS headers, emphasizing the need to run it behind a VPN or on an internal, non-publicly accessible server for security.

Postgres

Postgres

主な機能

ユースケース

主な機能

ユースケース