dltHub
Python-native embedded ELT library built from the ground up for LLMs and agents.
About
dlt is an open-source Python library for declarative data loading from APIs, databases, and files, built by dltHub. Unlike platform-style ELT tools, it is a library you pip install into your own code: it infers schemas automatically, handles schema evolution, and supports destinations including warehouses, vector databases like Weaviate, LanceDB, and Qdrant, and Delta Lake or Iceberg.
The project has leaned hard into the AI workflow: its docs are built for LLM consumption, the dltHub AI Workbench installs project rules and a secrets-management skill into your coding assistant, and an MCP server lets agents scaffold, debug, and validate pipelines without touching credentials directly. Its verified source catalog covers more than 5,000 sources.
What it can do
- Automatic schema inference and evolution
- Embedded library model, no separate infrastructure
- 60+ verified connectors and 5,000+ sources
- Vector database destinations for RAG pipelines
- LLM-native docs, MCP server, and AI Workbench
Setup
pip install dlt, then dlt init <source> <destination> to scaffold a pipeline you customize in Python.
Prerequisites
Python; API credentials stored in a local secrets.toml file.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 90; Trending score 80/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.