dst
Active·★ 16·Apache-2.0·Updated 2026-10-22
dst (data serve tool) is an open-source semantic layer and MCP server that connects AI tools like Claude, ChatGPT, and Copilot to your data warehouse, ensuring data accuracy and governance. It provides a controlled interface for AI to answer plain language questions with grounded numbers, SQL, confidence grades, and attached receipts, bringing engineering best practices to the data-to-AI lifecycle.
#Semantic Layer#Data Governance#AI Data Integration#Data Warehouse#SQL Generation#Data Testing#MLOps#Observability
01
Features
01Governed Semantic Layer: Provides a single, controlled interface between AI and data warehouses, ensuring AI answers are based on defined business logic.
02Automated Data Testing & Validation: Every approved answer becomes a regression test, running on every deploy to prevent silent data inaccuracies.
03Transparent Data Answers with Receipts: Every AI-generated data answer includes the producing SQL, confidence, and a verifiable receipt, enhancing trust and auditability.
04Engineering Best Practices for Data Serving: Incorporates software engineering principles like version control, CI/CD, and audit trails for managing data serving to AI.
05Typed Serving & Clarification: Instead of guessing, dst uses typed decisions to compile SQL and can ask for clarification when a word is ambiguous.
02
Compatibility
Python
Runtime
Verified via docs
Docker
Runtime
Verified via docs
Snowflake
Data Warehouse
Verified via docs
BigQuery
Data Warehouse
Verified via docs
Postgres
Data Warehouse
Verified via docs
DuckDB
Data Warehouse
Verified via docs
03
Quick start
1
$ pip install dst-core
2
$ dst init analytics --warehouse demo --yes
3
$ cd analytics
4
$ dst dev
04
Use cases
↳Providing accurate and governed data answers to conversational AI agents (e.g., ChatGPT, Claude) from enterprise data warehouses.
↳Preventing AI models from generating incorrect or speculative data analysis by grounding their responses in defined business logic and validated SQL.
↳Implementing a robust data governance framework for AI applications, ensuring all data access and interpretation are auditable and compliant.
↳Automating regression testing for data definitions and metrics, ensuring that changes to the semantic layer do not silently break existing data interpretations.
↳Developing internal AI agents or copilots that require reliable and verifiable access to an organization's analytical data.
05
Alternatives
orionbelt-semantic-layer★ 101
Open-source Semantic Sidecar for AI, analytics, and governed data systems. Compiles declarative YAML models into optimized SQL, semantic context, KPIs, and DQ rules.
publisher★ 118
Publisher is the open-source analytics engine for Malloy. It lets you define data models once — and use them everywhere.
lakevision★ 52
Lakevision is a tool which provides insights into your Apache Iceberg based Data Lakehouse.
awesome-agent-native-services★ 325
A curated list of agent-native services and infrastructure for AI agents: email, browsers, memory, sandboxes, payments, and MCP tools. Includes selection criteria and onboarding guides.
orionbelt-analytics★ 53
Ontology-based MCP server that analyzes database schemas (PostgreSQL, Snowflake, ClickHouse, Dremio) and generates RDF/OWL ontologies with SQL mappings for fan-trap-free Text-to-SQL.
Comments
Log in to leave a comment
No comments yet. Be the first!