I have no idea whether this is useful or not. But we built a BaseStore implementation that backs LangGraph long-term memory with Lithtrix, an external agent identity/memory API. Posting for feedback, not asking anyone to adopt it — we have zero users on this outside our own testing.
What it is: lithtrix-langgraph on PyPI implements batch/abatch over Lithtrix’s existing /v1/memory REST — get/put/delete/search/list-namespaces, values capped at 512 KiB, semantic search on the query param. No new Lithtrix endpoints, no fork of LangGraph.
What it isn’t: a general Lithtrix client. The package exports one thing — LithtrixStore — and nothing else. It doesn’t wrap Lithtrix’s other API surface (search, commons, agent directory, etc.); if you want that, you’d call the REST API directly.
Numbers: none to report on real usage yet — this is a fresh package, self-tested. The example in the docs was validated against the live production API, most recently by two independent, zero-assistance cold-run tests (an agent with no project context, following only the published PyPI page).
Known constraint: supports LangGraph 1.2.x (>=1.2.9,<1.3); custom BaseStore is still evolving upstream, so we pin within the minor we verified rather than claiming every future major.
Docs: LangGraph Integration - Lithtrix
This is a working reference, not a recommendation — happy to answer questions about the implementation or hear where theBaseStorecontract assumptions break.