// the find
dlt-hub/verified-sources
Contribute to dlt verified sources 🔥
A collection of pre-built dlt (data load tool) pipeline sources for loading data from APIs like Stripe, HubSpot, Salesforce, Kafka, and 25+ others into your data warehouse. You pull a source into your project via `dlt init <source>`, then customize it. Aimed at data engineers who want a working starting point rather than writing every API integration from scratch.
The `dlt init` injection model is the right call — you get the actual source code in your project to modify, not a black-box dependency you're locked into. The pg_replication source (logical replication via pgoutput) is genuinely useful and hard to implement correctly yourself. Source isolation is clean: each source has its own requirements.txt, so you're not dragging in every SDK at once. CI runs tests against real destinations (not mocks), which is how you actually catch breakage.
113 stars for the verified-sources repo is low signal given dlt itself has thousands — suggests people use the framework but don't contribute back here, which means quality and freshness of individual sources varies a lot and you can't assume any given source is production-tested. The explicit note that new sources won't be accepted if they 'can be implemented via REST API or with vibe coding' is a maintenance signal: the bar for inclusion was lowered in the past, leaving sources of uneven depth. No versioning or changelog per source — when a source changes, you won't know what broke unless you diff manually. Some sources (pokemon, chess) are clearly tutorial filler, not production connectors, which muddies expectations about what 'verified' actually means.