Mintplex-Labs/anything-llm
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/anything-llm.git synced 2026-07-21 01:25:28 -04:00
Code Issues 328 Packages Projects Releases 10 Wiki Activity

[GH-ISSUE #3679] [FEAT]: Agent max limits to prevent overthinking #2374

New Issue
Open
opened 2026-02-22 18:29:24 -05:00 by yindo · 0 comments
yindo commented 2026-02-22 18:29:24 -05:00
Owner
Copy Link
Copy Source

Originally created by @timothycarambat on GitHub (Apr 18, 2025).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3679

Add a way for the user in some way to specify hard caps on tool calls or other possible "overthinking" during agent execution, so that runaway calls do not result in surprise costs for some models or even lower-end models that hyper-focus on tool calls.

Hey @timothycarambat, thanks for the reply and for digging into the issue! I tested the agent with Claude 3.5 Sonnet and Gemini 2.5 Flash, and both worked well, finishing tasks quickly without burning through tokens or getting stuck in loops.

It does seem like 3.7’s behavior is the main culprit. Your idea of a tool iteration cap sounds like a solid fix to stop this kind of overuse with models like 3.7. Maybe something like Gemini’s thinking budget could also work here?

Originally posted by @LinusAurel in #3675

Originally created by @timothycarambat on GitHub (Apr 18, 2025). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3679 Add a way for the user in some way to specify hard caps on tool calls or other possible "overthinking" during agent execution, so that runaway calls do not result in surprise costs for some models or even lower-end models that hyper-focus on tool calls. > Hey @timothycarambat, thanks for the reply and for digging into the issue! I tested the agent with Claude 3.5 Sonnet and Gemini 2.5 Flash, and both worked well, finishing tasks quickly without burning through tokens or getting stuck in loops. > > It does seem like 3.7’s behavior is the main culprit. Your idea of a tool iteration cap sounds like a solid fix to stop this kind of overuse with models like 3.7. Maybe something like [Gemini’s thinking budget](https://ai.google.dev/gemini-api/docs/thinking) could also work here? _Originally posted by @LinusAurel in [#3675](https://github.com/Mintplex-Labs/anything-llm/issues/3675#issuecomment-2816103426)_
yindo changed title from [FEAT]: Agent `max` limits to prevent overthinking to [GH-ISSUE #3679] [FEAT]: Agent `max` limits to prevent overthinking 2026-06-05 14:46:11 -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
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#2374
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?