CentralMind Gateway: AI-First Data Gateway
🛸 Introduction
🚀 Interactive Demo via GitHub Codespaces
AI agents and LLM-powered applications need fast, secure access to data, but traditional APIs and databases aren't built for this purpose. We're building an API layer that automatically generates secure, LLM-optimized APIs for your structured data.
Our solution:
- Filters out PII and sensitive data to ensure compliance with GDPR, CPRA, SOC 2, and other regulations
- Adds traceability and auditing capabilities, ensuring AI applications aren't black boxes and security teams maintain control
- Optimizes for AI workloads, supporting Model Context Protocol (MCP) with enhanced meta information to help AI agents understand APIs, along with built-in caching and security features
Our primary users are companies deploying AI agents for customer support, analytics, where they need models to access the data without direct SQL access to databases elemenating security, compliance and peformance risks.
Features
- ⚡ Automatic API Generation – Creates APIs automatically using LLM based on table schema and sampled data
- 🗄️ Structured Database Support – Supports PostgreSQL, MySQL, ClickHouse, and Snowflake
- 🌍 Multiple Protocol Support – Provides APIs as REST or MCP Server including SSE mode
- 📜 API Documentation – Auto-generated Swagger documentation and OpenAPI 3.1.0 specification
- 🔒 PII Protection – Implements regex plugin or Microsoft Presidio plugin for PII and sensitive data redaction
- ⚡ Flexible Configuration – Easily extensible via YAML configuration and plugin system
- 🐳 Deployment Options – Run as a binary or Docker container with ready-to-use Helm chart
- 🤖 Multiple AI Providers Support - Support for OpenAI, Anthropic, Amazon Bedrock, Google Gemini & Google VertexAI
- 📦 Local & On-Premises – Support for self-hosted LLMs through configurable AI endpoints and models
- 🔑 Row-Level Security (RLS) – Fine-grained data access control using Lua scripts
- 🔐 Authentication Options – Built-in support for API keys and OAuth
- 👀 Comprehensive Monitoring – Integration with OpenTelemetry (OTel) for request tracking and audit trails
- 🏎️ Performance Optimization – Implements time-based and LRU caching strategies
How it Works
1. Connect & Discover
Gateway connects to your structured databases like PostgreSQL and automatically analyzes the schema and data samples to generate an optimized API structure based on your prompt. LLM is used only on discovery stage to produce API configuration. The tool uses AI Providers to generate the API configuration while ensuring security through PII detection.
2. Deploy
Gateway supports multiple deployment options from standalone binary, docker or Kubernetes. Check our launching guide for detailed instructions. The system uses YAML configuration and plugins for easy customization.
3. Use & Integrate
Access your data through REST APIs or Model Context Protocol (MCP) with built-in security features. Gateway seamlessly integrates with AI models and applications like LangChain, OpenAI and Claude Desktop using function calling or Cursor through MCP. You can also setup telemetry to local or remote destination in otel format.
Documentation
Getting Started
Additional Resources
How to Build
# Clone the repository
git clone https://github.com/centralmind/gateway.git
# Navigate to project directory
cd gateway
# Install dependencies
go mod download
# Build the project
go build . ```
## API Generation
Gateway uses LLM models to generate your API configuration. Follow these steps:
1. Create a database connection configuration file (`connection.yaml`):
```yaml type: postgres hosts:
* localhost user: 'your-database-user' password: 'your-database-password' database: 'your-database-name' port: 5432 ```
2. Choose one of our supported AI providers:
* [OpenAI](https://docs.centralmind.ai/providers/openai) and all OpenAI-compatible providers
* [Anthropic](https://docs.centralmind.ai/providers/anthropic)
* [Amazon Bedrock](https://docs.centralmind.ai/providers/bedrock)
* [Google Vertex AI (Anthropic)](https://docs.centralmind.ai/providers/anthropic-vertexai)
* [Google Gemini](https://docs.centralmind.ai/providers/gemini)
[Google Gemini](https://docs.centralmind.ai/providers/gemini) provides a generous **free tier**. You can obtain an API key by visiting Google AI Studio:
* [Google AI Studio](https://aistudio.google.com/apikey)
Once logged in, you can create an API key in the API section of AI Studio. The free tier includes a generous monthly token allocation, making it accessible for development and testing purposes.
Configure AI provider authorization. For Google Gemini, set an API key.
```bash export GEMINI_API_KEY='yourkey' ```
3. Run the discovery command:
```shell ./gateway discover \ \--ai-provider gemini \ \--config connection.yaml \ \--prompt "Generate for me awesome readonly API" ```
4. Monitor the generation process:
```shell INFO 🚀 API Discovery Process INFO Step 1: Read configs INFO ✅ Step 1 completed. Done.
INFO Step 2: Discover data INFO Discovered Tables: INFO - payment_dim: 3 columns, 39 rows INFO - fact_table: 9 columns, 1000000 rows INFO ✅ Step 2 completed. Done.
# Additional steps and output...
INFO ✅ All steps completed. Done.
INFO --- Execution Statistics --- INFO Total time taken: 1m10s INFO Tokens used: 16543 (Estimated cost: $0.0616) INFO Tables processed: 6 INFO API methods created: 18 INFO Total number of columns with PII data: 2 ```
5. Review the generated configuration in `gateway.yaml`:
```yaml api: name: Awesome Readonly API description: '' version: '1.0' database: type: postgres connection: YOUR_CONNECTION_INFO tables: \- name: payment_dim columns: # Table columns endpoints: \- http_method: GET http_path: /some_path mcp_method: some_method summary: Some readable summary description: 'Some description' query: SQL Query with params params: # Query parameters ```
## Running the API
### Run locally
```shell ./gateway start --config gateway.yaml rest ```
### Docker Compose
```shell docker compose -f ./example/simple/docker-compose.yml up ```
### MCP Protocol Integration
Gateway implements the MCP protocol for seamless integration with Claude and other tools. For detailed setup instructions, see our [Claude integration guide](https://docs.centralmind.ai/docs/content/integration/claude-desktop/).
1. Build the gateway binary:
```shell go build . ```
2. Configure Claude Desktop tool configuration:
```json { "mcpServers": { "gateway": { "command": "PATH_TO_GATEWAY_BINARY", "args": ["start", "--config", "PATH_TO_GATEWAY_YAML_CONFIG", "mcp-stdio"] } } } ```
## Roadmap
It is always subject to change, and the roadmap will highly depend on user feedback. At this moment, we are planning the following features:
#### Database and Connectivity
* 🗄️ **Extended Database Integrations** - Redshift, S3 (Iceberg and Parquet), Oracle DB, Microsoft SQL Server, Elasticsearch
* 🔑 **SSH tunneling** - ability to use jumphost or ssh bastion to tunnel connections
#### Enhanced Functionality
* 🔍 **Advanced Query Capabilities** - Complex filtering syntax and Aggregation functions as parameters
* 🔐 **Enhanced MCP Security** - API key and OAuth authentication
#### Platform Improvements
* 📦 **Schema Management** - Automated schema evolution and API versioning
* 🚦 **Advanced Traffic Management** - Intelligent rate limiting, Request throttling
* ✍️ **Write Operations Support** - Insert, Update operations
