Mintplex-Labs/anything-llm
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/anything-llm.git synced 2026-07-19 14:13:35 -04:00
Code Issues 328 Packages Projects Releases 10 Wiki Activity

[GH-ISSUE #3525] [BUG]: Agent inconsistency when processing uploaded documents #2272

New Issue
Closed
opened 2026-02-22 18:28:56 -05:00 by yindo · 1 comment
yindo commented 2026-02-22 18:28:56 -05:00
Owner
Copy Link
Copy Source

Originally created by @MiguelGarcia-Paccar on GitHub (Mar 24, 2025).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3525

How are you running AnythingLLM?

AnythingLLM desktop app

What happened?

I'm attempting to use agents to fetch and process information from 2 documents. Issue is, when asking the agent if it acknowledges then, it will output yes, and right afterwards, it will say that it doesn't. For reference, I did not change anything in between prompts. Any ideas? or is this a real bug?

Image

Image

Image

Image

This are the embedded documents:

Image

agent config

Image

Are there known steps to reproduce?

No response

Originally created by @MiguelGarcia-Paccar on GitHub (Mar 24, 2025). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3525 ### How are you running AnythingLLM? AnythingLLM desktop app ### What happened? I'm attempting to use agents to fetch and process information from 2 documents. Issue is, when asking the agent if it acknowledges then, it will output yes, and right afterwards, it will say that it doesn't. For reference, I did not change anything in between prompts. Any ideas? or is this a real bug? ![Image](https://github.com/user-attachments/assets/f56adc8a-22c3-4da7-9bbd-16c4d30b67fd) ![Image](https://github.com/user-attachments/assets/f59bedf3-7f12-4d07-9734-9c9bb4cd20cf) ![Image](https://github.com/user-attachments/assets/117faa38-e809-4efb-97cd-4d8e1b42132f) ![Image](https://github.com/user-attachments/assets/9758d9bf-adaa-4a0e-88b6-43db239c32c1) This are the embedded documents: ![Image](https://github.com/user-attachments/assets/ec06a5e2-460d-4e35-9862-d3ca954699a6) agent config ![Image](https://github.com/user-attachments/assets/a58245ee-11d0-4ecf-96ad-2fb787ddbee8) ### Are there known steps to reproduce? _No response_
yindo added the possible bug label 2026-02-22 18:28:56 -05:00
yindo closed this issue 2026-02-22 18:28:56 -05:00
yindo commented 2026-02-22 18:28:56 -05:00
Author
Owner
Copy Link
Copy Source

@timothycarambat commented on GitHub (Mar 24, 2025):

You for sure have documents embedded, and its clear the model can see it when the agent tool is called (expected). However, what you are seeing is simply a model hallucination - that is all. There is no bug it is simply the model not calling the tool you expect it to see so it responds as if it cannot see the documents, because it cannot.

https://docs.anythingllm.com/agent-not-using-tools

Fundamentally, agent capabilities come down to the abilities of a model to call JSON tools. Sometimes context can impact the models abilities more than others as it varies from model to model. Sometimes some models get far too distracted with any history + tools calls and need to be /reset in the chat to get better results. Additionally, more active tools = more context to process, and the LLM gets lost in the context!

@timothycarambat commented on GitHub (Mar 24, 2025): You for sure have documents embedded, and its clear the model can see it when the agent tool is called (expected). However, what you are seeing is simply a model hallucination - that is all. There is no bug it is simply the model not calling the tool you expect it to see so it responds as if it cannot see the documents, because it cannot. https://docs.anythingllm.com/agent-not-using-tools Fundamentally, agent capabilities come down to the abilities of a model to call JSON tools. Sometimes context can impact the models abilities more than others as it varies from model to model. Sometimes some models get far too distracted with any history + tools calls and need to be `/reset` in the chat to get better results. Additionally, more active tools = more context to process, and the LLM gets lost in the context!
yindo changed title from [BUG]: Agent inconsistency when processing uploaded documents to [GH-ISSUE #3525] [BUG]: Agent inconsistency when processing uploaded documents 2026-06-05 14:45:34 -04:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
Desktop
Docker
Integration Request
Integration Request
OS: Linux
OS: Mobile
OS: Windows
UI/UX
blocked
bug
bug
core-team-only
documentation
duplicate
embed-widget
enhancement
feature request
github_actions
good first issue
investigating
needs info / can't replicate
possible bug
pull-request
Mirrored from GitHub Pull Request
 question
stage: specifications
wontfix
No Label possible bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#2272
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?