How to get all threads from langgraph so I can display all chats on the side panel ?

[tyadav7]

Hey,

I want to get all the conversations that happened for a particular user. So that I can display it on the side panel. Is there a way to fetch all threads from the underlying memory of the agent?

[hinthornw]

Usually something like:

client.threads.search(metadata={"user_id": ...})

https://langchain-ai.github.io/langgraph/cloud/reference/sdk/python_sdk_ref/#langgraph_sdk.client.ThreadsClient.delete

[tyadav7]

I am not using langgraphsdk client.

I am talking about the CompileStateGraph since each CompileStateGraph has a short term memory persisted in a Checkpointer is there no way to get all conversations states on the CompileStateGraph and then filter the threads based on a user?

[hinthornw]

Then checkpointer.list()?

https://langchain-ai.github.io/langgraph/reference/checkpoints/#langgraph.checkpoint.base.BaseCheckpointSaver.list

But I think if you want to manage a list of threads and related checkpoints that'd be a separate table you create

[Freezaa9]

Hi, @hinthornw
I'm looking to achieve the same but list require a config with a thread_id so we can only filter on a given thread.
To get all thread from a user we would need to filter on all thread.

[hinthornw]

Yeah if you're rolling your own server then you'll have to create a separate threads table and associate threads with users yourself.

That's an abstraction above the checkpointer today

[EzzatEsam]

It would be nice if we got this as a feature for BaseCheckpointSaver interface to fetch checkpoints without having to rely on thread id (possibly with other filters for config too).
I went with creating my own way of keeping track of threads & user association (Separate table).

However, if you just wanted to get all thread ids, you can implement it yourself. Langgraph uses a table called checkpoints.

if you were using sqlite you may do st like this

import aiosqlite

async def get_distinct_thread_ids(db_path: str)->list[str]:
    async with aiosqlite.connect(db_path) as db:
        async with db.execute("SELECT DISTINCT thread_id FROM checkpoints") as cursor:
            rows = await cursor.fetchall()
            return [row[0] for row in rows]