valv - Database Queries Without Writing SQL

Let AI agents query your database safely—no SQL, just validated queries.

Unclaimed

Added Jul 2026

valv.sh

Mcp DataFreemcp-data

Visit Valv Ts Library

Share: X in LinkedIn

Add your screenshot here

Image or video shown in this spot

Get Verified · $10 lifetime

What is Valv Ts Library?

valv is a TypeScript library that enables AI agents to safely query databases without ever writing raw SQL. It provides secure, policy-based access control where every query is validated against your schema server-side, works with any database (Postgres, MySQL, SQLite, ClickHouse), and integrates with Vercel AI SDK or MCP. It's designed for developers who want to give agents real database access while maintaining security and controlling token usage.

Explore Valv Ts Library

Need help implementing valv - Database Queries Without Writing SQL?

Find verified specialists who work with valv - Database Queries Without Writing SQL

Browse specialists

Key Features of Valv Ts Library

Secure-by-construction query validation
Server-side schema introspection and re-checking
Policy-based access control in TypeScript
Deny-by-default permission scoping
Token-aware result capping
Support for Postgres, MySQL, SQLite, ClickHouse
Integration with Vercel AI SDK and MCP
Real analytics queries (filters, aggregates, time-series, top-N)
Per-caller policy enforcement

Who Should Use Valv Ts Library?

Enable agents to perform analytics queries on databases

Provide secure database access for multi-tenant applications

Implement row and column-level access control for agents

Allow agents to answer complex data questions with aggregations and filtering

Manage agent token usage when querying large datasets

Valv Ts Library: Pros & Cons

✓Pros

Agents never write raw SQL, eliminating SQL injection risks
Policies are code-based and audit-friendly
Supports multiple database backends without code changes
Token-aware capping prevents context overflow and budget overruns
Same agent can return different results based on caller's permissions
Open source and free
Works with existing agent frameworks (Vercel AI SDK, MCP)

Frequently Asked Questions about Valv Ts Library

How does valv keep database queries secure?

valv prevents the model from ever writing SQL. Instead, agents send structured, validated queries that are rebuilt and re-checked against your schema on the server, eliminating SQL injection risks.

What databases does valv support?

valv works with Postgres, MySQL, SQLite, and ClickHouse. It introspects your live schema automatically upon connection.

Can valv control what data different users see?

Yes. valv applies deny-by-default policies written in TypeScript that scope every row and column to the current caller, so the same question returns different results based on permissions.

How does valv prevent agents from consuming too many tokens?

valv is token-aware and caps how much data comes back from queries, preventing wide-open queries from flooding the model's context or burning your budget.

Can valv integrate with existing AI frameworks?

Yes. valv works with the Vercel AI SDK and can be exposed over MCP, dropping into your existing stack.