Contributing an Aerospike persistence integration to LangGraph

OSS Product Help LangGraph
,
1

Hi LangGraph team,

Our team at Aerospike is building an Aerospike integration for LangGraph persistence and wanted to check on the best way to contribute it upstream.

So far, We have implemented:

  • A custom checkpointer (BaseCheckpointSaver)

  • A custom memory / store backend (BaseStore)

Before opening a PR, we wanted to align on a few things:

  • Structure: Should DB integrations live inside the LangGraph repo or as external/community packages? Is there a preferred pattern to follow?

  • Process: Is it better to start with a design discussion or open a draft PR? Any existing integrations you’d recommend referencing?

Happy to share a repo or design doc if helpful.
Thank you!
Aerospike Team

1 Like
2

Hello, thanks for sharing this, this is exciting!

Typically for third party integrations:

  • The library is maintained outside of LangChain monorepos (e.g., it could remain in a Github repo your org controls)
  • We document integrations centrally in LangChain’s docs repo.

We have less organization around LangGraph checkpointer and memory store integrations (some are listed on the memory page here and here). We will likely structure these in a dedicated page in the future.

For now I’d suggest:

3

Thank you for the guidance!

We’ve published the integration as standalone packages with usage documented in the READMEs, and we’ve opened a docs issue:

https://github.com/langchain-ai/docs/issues/2131

Thank you again, looking forward to contributing!