graphjin: a governed GraphQL and MCP gateway that lets AI agents securely query data systems and source code
graphjin: a governed GraphQL and MCP gateway that lets AI agents securely query data systems and source code
What it solves
GraphJin provides a unified, governed interface for AI agents to interact with a company's existing data systems. It prevents agents from needing raw database credentials or guessing schema structures, instead offering a secure, auditable layer that translates agent requests into optimized queries across databases, warehouses, files, and source code.
How it works
GraphJin acts as a compiler and runtime that maps various data sources—including PostgreSQL, MySQL, MongoDB, Snowflake, S3, and local files—into a single GraphQL and Model Context Protocol (MCP) surface. It automatically discovers schemas and relationships, allowing agents to discover capabilities before acting. It includes a feature called CodeSQL, which indexes source code using tree-sitter to make code searchable and queryable like a database, enabling agents to link production data to the actual code that operates it.
Who it’s for
It is designed for developers building AI agents that need secure, read-only or read-write access to complex enterprise data environments, as well as developers who want a high-performance GraphQL-to-database compiler for their applications.
Highlights
- Unified Data Surface: Accesses operational databases, warehouses, object stores, and source code through one interface.
- Built-in Agentic Loop: Includes a server-side agent that can run discovery and return evidence-backed answers via a single API call.
- CodeSQL: Indexes source code to allow agents to query symbols, references, and imports alongside database tables.
- Governed Access: Implements query allow-lists, read-only boundaries, and policy-aware MCP tools to ensure security.
- Durable Memory: Supports saved queries, fragments, and workflows with cursor-backed watches for real-time updates.
- Broad Compatibility: Supports a wide array of backends including PostgreSQL, MySQL, MongoDB, SQLite, Oracle, MSSQL, Snowflake, Redshift, BigQuery, and S3/GCS.
Sources
- undefineddosco/graphjin