gateway
Basic Information
CentralMind Gateway is a server and toolkit that exposes structured databases to AI agents and LLM-powered applications by automatically generating secure, AI-optimized APIs. It can run as a standalone binary, in Docker, or on Kubernetes and serves data over REST/OpenAPI and the Model Context Protocol (MCP) including SSE mode. The tool discovers database schema and sample data, uses LLMs at discovery to produce API configuration, and offers direct/raw SQL endpoints. It aims to make it fast and simple for developers to provide model-aware, auditable access to databases while applying security and privacy controls during API generation and runtime.
Links
Stars
447
Github Repository
App Details
Features
Automatic API generation driven by LLM-assisted discovery of schema and samples. Support for many databases including PostgreSQL, MySQL, ClickHouse, Snowflake, MSSQL, BigQuery, Oracle, SQLite and Elasticsearch. Multi-protocol exposure via REST/OpenAPI and MCP with SSE. Authentication options including API keys and OAuth. PII protection via regex or Microsoft Presidio plugins and row-level security via Lua scripts. Observability integration with OpenTelemetry for tracing and auditing. Caching strategies including time-based and LRU. Extensible configuration with YAML and a plugin system. Multi-provider AI support for OpenAI, Anthropic, Amazon Bedrock, Google Gemini and Vertex AI. Deployable as Docker image, binary or on Kubernetes and emits Swagger/OpenAPI docs.
Use Cases
Gateway shortens the time required to give AI agents secure, documented access to live data by automating API creation and providing production features like access controls, auditing and PII redaction. It enables developers and data teams to chat with data warehouses, empower agents with remote function or tool calling, and integrate with frameworks like LangChain or Claude via MCP. Built-in telemetry and traceability help security and compliance teams monitor usage. The runtime supports self-hosted models and multiple cloud AI providers, so teams can choose deployment and model strategies that fit privacy and cost requirements while keeping APIs consistent and optimized for LLM workloads.
-
$$$$ -
$$$$