DatI (Data Intelligence) is a database semantic gateway (Semantic Gateway) designed for AI agents. Its core goal is to transform common databases into secure, semantically rich MCP tools for agents or any MCP host to call.
Main capabilities
- Multi-database support: Supports MySQL, PostgreSQL, ClickHouse, Doris, and other relational and analytical databases.
- Semantic enhancement: Supports business terms, column aliases, and automatic enumeration dictionary extraction; combined with semantic search, it helps models understand business terms and find the correct tables.
- Flexible integration: Based on standard MCP (Streamable HTTP), it can be easily integrated into existing agents or workflows.
- Easy MCP publishing: Provides out-of-the-box built-in tools (schema viewing, SQL execution) and parameterized SQL tools, enabling MCP services to be published with zero additional deployment.
- Security and governance: Centrally manages database credentials and supports user-level permissions and data scope isolation.
Usage
- Local development: After startup, open http://localhost:8085 , and register with the username admin to obtain the super administrator role. For production environment tuning and external database setup, see the deployment guide.
- Demo instance:https://dati-demo.zhangyimin.me . The demo instance is preloaded with sample data and reset periodically. The data and configuration reference AdventureWorks samples and can be automated through the dati-ops skill.
- Agent skills: Follows the Agent Skills Open Standard and can be automatically discovered within the repository. Among them, dati-ops is a user skill that configures and operates the platform through the HTTP API (data sources, topics and terms, MCP services), embeds openapi.json and query tools, can be distributed independently, and is connected within the repository through .agents/skills/dati-ops/; e2e-tester is a developer skill used for E2E HTTP integration testing and API behavior verification (test cases are in e2e-tests/test-cases/).
Tech stack
- Backend: Spring Boot 3.5.x + Java 21 + JPA
- Frontend: Vue 3 + TypeScript + Vite + Element Plus + TailwindCSS 4
- Database: H2 (development) / MySQL / PostgreSQL (production)
- Search engine: Elasticsearch (semantic retrieval)
Documentation and resources
- Local development guide: environment setup, backend/frontend startup, and project conventions
- Server deployment guide: architecture overview, Docker Compose deployment, and operations
- Architecture overview: high-level architecture, module index, and key design conventions
- User guide: docs/user-guide (VitePress bilingual site)
- API specification: docs/api/openapi.json (for the E2E testing toolchain)
- AI coding assistant guide: AGENTS.md and .agents/rules/ (backend, frontend, and design system rules)
Example scenarios
- Enterprise BI and retail analytics (AdventureWorks DW): star schema Text-to-SQL, metric governance, multi-table joins, and parameterized acceleration.
- Family finance assistant: multi-user collaborative bookkeeping, parameterized permission control, transparent SQL queries, and self-healing agent workflows.
Applicable scenarios
- Conversational data analysis (NL2SQL): connect to business databases and support NL2SQL analysis workflows with semantic metadata and built-in tools.
- Lightweight application development: wrap databasesas MCP services, enabling agents to query and update data through conversation and build lightweight applications.
Limitations The materials do not provide information about fees, pricing, platform compatibility, effectiveness metrics, or specific performance data; other unmentioned limitations are not provided at this time.
