Can archon-memory-core be used as a LangGraph Store backend?

Indirectly — you can call archon-memory-core from inside a LangGraph node or tool. There isn't a drop-in BaseStore adapter yet, but the surface area is small enough to write one in an afternoon.

archon-memory-core vs LangGraph Store

Q: Does LangGraph Store do long-term memory?

Yes — LangGraph Store is a key-value store with namespaced scopes (user, thread, etc.) and optional vector search. It's a primitive you build on; retrieval policy is your responsibility.

Q: Do I need archon-memory-core if I'm using LangGraph?

Not necessarily. If your memory is simple key-value with vector search, LangGraph Store is enough. If you need persistence classes, priority scoring, and contradiction resolution out of the box, archon-memory-core gives you those primitives.

LangGraph Store is LangChain's built-in memory primitive — a namespaced key-value store with optional vector search. archon-memory-core is a framework-agnostic memory library with opinionated contradiction handling. One is a primitive; the other is a policy.

Short answer: If you're deep in LangGraph and your memory is "save a blob, retrieve it later," LangGraph Store is already there. If you want persistence classes, priority scoring, and retrieval-time contradiction resolution without writing them yourself, use archon-memory-core — either standalone or called from a LangGraph node.

What each one actually is

LangGraph Store is a pluggable interface with an in-memory default and Postgres / SQLite implementations. You get put, get, search against namespaced scopes. Vector search is optional. It's a durable bag you fill however you like — the retrieval policy and conflict handling are up to you.

archon-memory-core has a narrower API but makes opinionated decisions. Every memory has a persistence class (ephemeral, session, long-term, canonical) and an optional priority. Retrieval ranks by a combination of embedding similarity, persistence, and priority. Contradictions coexist in storage and get sorted at query time.

Feature comparison

Capability	LangGraph Store	archon-memory-core
License	MIT	Apache 2.0
Install	Part of LangGraph	`pip install archon-memory-core`
Framework coupling	Tightly integrated with LangGraph	Framework-agnostic
API shape	KV store (put / get / search)	Memory client (write / retrieve)
Persistence classes	Not built-in	Built-in (ephemeral → canonical)
Priority scoring	DIY	Built-in
Contradiction handling	DIY	Retrieval-time, deterministic
Benchmarks	None published specifically	AMB v2.3 (99.2% top-1)

When to pick LangGraph Store

You're already deep in LangGraph and want zero new dependencies.
Your memory is simple: save a JSON blob, look it up later, maybe vector search.
You want full control over what gets written and how it's retrieved.

When to pick archon-memory-core

You want persistence class and priority as first-class concepts instead of reinventing them.
You want contradiction resolution that doesn't require you to write ranking logic yourself.
You want something that works unchanged if you swap LangGraph for CrewAI, Pydantic AI, or a custom stack.
You care about the AMB v2.3 preregistered number (99.2% top-1 on contradiction retrieval).

Using them together

LangGraph Store is great for checkpointing, thread state, and anything that wants LangGraph's graph-scoped namespacing. Call archon-memory-core from inside a node when you need durable user-level memory with contradiction handling. A tiny adapter that implements LangGraph's BaseStore on top of archon-memory-core is on the roadmap — until then, direct invocation from a node works fine.

FAQ

Does LangGraph Store do long-term memory?

Yes — it's a namespaced key-value store with optional vector search. Retrieval policy is your responsibility.

Do I need archon-memory-core if I'm using LangGraph?

Not if your memory is simple KV + vector. Yes if you want persistence classes, priority, and contradiction resolution without writing them yourself.

Can archon-memory-core back LangGraph Store?

Indirectly today — call it from inside a node. A BaseStore adapter is on the roadmap.

Which is better tested?

LangGraph Store benefits from LangChain's user base. archon-memory-core ships with AMB v2.3, a preregistered benchmark measuring contradiction retrieval directly.

View archon-memory-core on GitHub