Memory Configuration

[jakub-tomanik]

What happened?

I've been running local docker based installation of LibreChat for many months and never had any major issues and have been using it on a daily basis. I'm updating the image at most every 2-3 weeks.

Yesterday I have updated my docker setup with the usual docker compose pull and I was not able to use it since. For every model that I have configured I'm getting the message:

"Something went wrong. Here's the specific error message we encountered: An error occurred while processing your request. Please contact the Admin."

And it happens for both old (when resuming) and new conversations.

Docker logs show:

error: [handleAbortError] AI response error; aborting request: Cannot read properties of null (reading 'model')

To be more speciffic: I'm using only OpenAI, Anthropic and Google endpoints.
API keys are in .env file and the only model override there is for Google with

GOOGLE_MODELS=gemini-2.5-pro,gemini-2.5-flash,gemini-2.5-flash-lite-preview-06-17,gemini-2.0-flash,gemini-2.0-flash-lite

I don't have any agents configured

Version Information

docker images | grep librechat
librechat/librechat-dev                          latest       17ad90003774   23 hours ago    1.11GB
ghcr.io/danny-avila/librechat-rag-api-dev-lite   latest       0b5a9e43bafe   13 days ago     1.44GB

Steps to Reproduce

1. Update docker image to the librechat-dev:latest
2. Log in
3. Try to use any model

What browsers are you seeing the problem on?

No response

Relevant log output

from docker console:

LibreChat         | 2025-07-03T16:38:59.149Z debug: [BaseClient] Loading history:
LibreChat         | {
LibreChat         |   conversationId: "83e6de9d-4c25-4be7-b2e1-fa2b0dbc3b41",
LibreChat         |   parentMessageId: "00000000-0000-0000-0000-000000000000",
LibreChat         | }
LibreChat         | 2025-07-03T16:39:01.120Z debug: [BaseClient] Context Count (1/2)
LibreChat         | {
LibreChat         |   remainingContextTokens: 7357.3,
LibreChat         |   maxContextTokens: 7368.3,
LibreChat         | }
LibreChat         | 2025-07-03T16:39:01.124Z debug: [BaseClient] Context Count (2/2)
LibreChat         | {
LibreChat         |   remainingContextTokens: 7357.3,
LibreChat         |   maxContextTokens: 7368.3,
LibreChat         | }
LibreChat         | 2025-07-03T16:39:01.129Z debug: [BaseClient] tokenCountMap:
LibreChat         | {
LibreChat         |   c122a56b-e5ae-4a3f-9b29-9a6a699ac69d: 8,
LibreChat         | }
LibreChat         | 2025-07-03T16:39:01.136Z debug: [BaseClient]
LibreChat         | {
LibreChat         |   promptTokens: 11,
LibreChat         |   remainingContextTokens: 7357.3,
LibreChat         |   payloadSize: 1,
LibreChat         |   maxContextTokens: 7368.3,
LibreChat         | }
LibreChat         | 2025-07-03T16:39:01.188Z error: [handleAbortError] AI response error; aborting request: Cannot read properties of null (reading 'model')
LibreChat         | 2025-07-03T16:39:01.350Z debug: [AgentController] Performing cleanup
LibreChat         | 2025-07-03T16:39:01.352Z debug: [AgentController] Cleaning up abort controller
LibreChat         | 2025-07-03T16:39:01.356Z debug: [AgentController] Cleanup completed


from /logs/error-*.log


{"level":"error","message":"[handleAbortError] AI response error; aborting request: Cannot read properties of null (reading 'model')","stack":"TypeError: Cannot read properties of null (reading 'model')\n    at initializeAgent (/a>
{"level":"error","message":"[handleAbortError] AI response error; aborting request: Cannot read properties of null (reading 'model')","stack":"TypeError: Cannot read properties of null (reading 'model')\n    at initializeAgent (/a>
{"level":"error","message":"[handleAbortError] AI response error; aborting request: Cannot read properties of null (reading 'model')","stack":"TypeError: Cannot read properties of null (reading 'model')\n    at initializeAgent (/a>
{"level":"error","message":"[handleAbortError] AI response error; aborting request: Cannot read properties of null (reading 'model')","stack":"TypeError: Cannot read properties of null (reading 'model')\n    at initializeAgent (/a>
{"level":"error","message":"[handleAbortError] AI response error; aborting request: Cannot read properties of null (reading 'model')","stack":"TypeError: Cannot read properties of null (reading 'model')\n    at initializeAgent (/a>

Screenshots

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[jakub-tomanik]

Turning off memory feature restores the basic functionality
Memory configuration was copied from the website:

# Memory configuration for user memories
memory:
  # (optional) Disable memory functionality
  disabled: false
  # (optional) Restrict memory keys to specific values to limit memory storage and improve consistency
#   validKeys: ["preferences", "work_info", "personal_info", "skills", "interests", "context"]
  # (optional) Maximum token limit for memory storage (not yet implemented for token counting)
#   tokenLimit: 10000
  # (optional) Enable personalization features (defaults to true if memory is configured)
  # When false, users will not see the Personalization tab in settings
  personalize: true
  # Memory agent configuration - either use an existing agent by ID or define inline
  messageWindowSize: 10
  agent:
    # Option 1: Use existing agent by ID
    id: "memory-specialist-001"

[danny-avila]

Hey thanks for the report. This makes sense, as the agent ID should be valid a supported provider/model configuration is added.

[jakub-tomanik]

I think I may have misread the https://www.librechat.ai/docs/features/memory#overview and got an impression that memory-specialist-001 is a built in memory agent. If this is not the case then technically this is not a bug but it's easy to misunderstood the docs

[jakub-tomanik]

To add to the confusion when I checked the website now at the top it has:

"Currently, memory only works when using agents."

so II would expect that even if misconfigured it should not affect normal conversations.

Anyways Danny I am a great fan and have been using this amazing tool everyday for months. Please decide yourself if this is a genuine bug or a problem with docs.

Anyways I decided to make a bug report as searching for this particular error didn't return any result. Just in case anyone one else starts to panic they have messed up the upgrade ;)

[danny-avila]

"Currently, memory only works when using agents."

This needs to be updated, as it now works with all chat options.

You can also configure memory as simply as this:

memory:
  tokenLimit: 2000
  agent:
    provider: "azureOpenAI"
    model: "gpt-4.1-mini"

Thanks for the kind words!

[akshayi1]

@danny-avila - thank you for the awesome work! I'm using Railway to deploy the app because I wanted to use it on my phone as well, and don't have a 24/7 running machine to keep the service up and running. I was curious about how to use Memory as well, but I think it doesn't work as I'd expect it to. It's not very clear, if the usage of a memory is tied to an agent or not. I added my memories from my ChatGPT profile into Librechat in the UI, but it doesn't look like the models I chat with can access (or even reference) those. Can you please guide me on how to have an agent-agnostic memory? I use Openrouter and would love for any model I'm using to know about the memories. Thanks again for the tree do us work!

[maxjacu]

Unable to access memories as well. Litellm is our custom endpoint.