AskDB is a self-hosted bridge that provides AI agents secure, sandboxed access to your MongoDB or PostgreSQL databases, ensuring fine-grained control over data visibility and comprehensive auditing.

How does AskDB protect my production data?

It clones your production data into isolated Docker containers for AI access, never the original. It also features robust query validation to prevent write operations and dangerous stages, ensuring read-only safety.

What databases are supported by AskDB?

AskDB currently supports both MongoDB and PostgreSQL databases. It provides a unified Model Context Protocol (MCP) tool surface across these different database engines.

Can I control which data fields AI agents can view?

Yes, AskDB offers granular field-level visibility control. You can toggle specific fields (including auto-detected PII like email or SSN) visible or hidden, with changes taking effect immediately without re-syncing.

Does AskDB track AI agent database interactions?

Absolutely. AskDB maintains a comprehensive audit trail, logging every MCP query with timestamps, execution details, target database, and row count for full transparency and compliance.

AskDB

Name: AskDB
Author: mgorabbani

•

Gives AI agents safe, sandboxed access to your databases, allowing fine-grained control over data visibility and comprehensive query auditing.

AskDB

•

Gives AI agents safe, sandboxed access to your databases, allowing fine-grained control over data visibility and comprehensive query auditing.

AskDB is a self-hosted bridge designed to securely connect your MongoDB or PostgreSQL databases with AI agents that communicate via the Model Context Protocol (MCP). It clones your production data into an isolated sandbox, letting you meticulously control which fields and collections AI agents can see, ensuring sensitive information is never exposed. This powerful tool provides a unified MCP endpoint for various AI tools like Claude, ChatGPT, and Cursor, complete with robust query validation, PII auto-detection, and an audit trail for every database interaction.

主な機能

01Sandbox isolation for production data with Docker containers.

02Comprehensive audit trail for every MCP query with timestamps and execution details.

03Field-level visibility control with PII auto-detection and immediate effect.

04Supports MongoDB and PostgreSQL databases.

05Robust query validation to allowlist read-only operations and block dangerous stages.

061 GitHub stars

ユースケース

01Securely connecting AI agents (e.g., Claude, ChatGPT, Cursor) to real-time production databases.

02Controlling and masking sensitive data (PII) from AI agents at a field level.

03Providing a unified and secure Model Context Protocol (MCP) endpoint for all AI tools to interact with diverse database engines.