LlamaHub is archived; long live the monorepo

What it does

LlamaHub was a community-maintained registry of data loaders, agent tools, and benchmark datasets for LLM applications. It provided connectors to ingest data from sources like Google Docs, Notion, Slack, and SQL databases, or to let agents read and write to third-party services. The project also included pre-built workflow templates called llama-packs and evaluation datasets for testing RAG pipelines.

The interesting bit

With LlamaIndex v0.10, the maintainers moved every integration—loaders, tools, packs, and datasets—into the core llama-index Python repository and archived this standalone project. The LlamaHub brand survives mainly as the llamahub.ai discovery portal, while the code itself is now frozen in read-only mode.

Key highlights

• Covers a wide surface area: document loaders, agent tools, reusable packs, and RAG evaluation datasets.
• Originally created by Jesse Zhang and donated to the LlamaIndex organization.
• Designed to work with both LlamaIndex and LangChain.
• Now read-only; all integrations were migrated into the core llama-index repository.
• The llamahub.ai site continues to catalog integrations from the core repo.

Caveats

• The repository is explicitly archived and no longer accepts updates.
• Contribution guides and setup instructions in the README are now historical artifacts.

Verdict

Worth browsing if you are maintaining legacy LlamaIndex code or curious about pre-v0.10 connectors. Skip it if you are starting fresh—head straight to the core llama-index repository instead.