Apply model failing

Discussion
1

I don’t know if it’s related to platform availability, but currently composer in Agent mode isn’t is failing to apply changes.

I’m not sure which model does the apply, but currently requests are failing to apply from both claude-3.5-sonnett and gpt-4o.

Version: 0.44.11
VSCode Version: 1.93.1
Commit: fe574d0820377383143b2ea26aa6ae28b3425220
Date: 2025-01-03T07:59:06.361Z
Electron: 30.5.1
Chromium: 124.0.6367.243
Node.js: 20.16.0
V8: 12.4.254.20-electron.0
OS: Windows_NT x64 10.0.26120

claude-3.5-sonnet:

I apologize, but it seems the apply model is consistently failing to make changes to the file. This might be due to some underlying issues with the file state or the model’s ability to parse the current content.

gpt-4o:

I apologize for the difficulty with the apply model. It seems to be having trouble making the changes. Let me know if you’d like me to try a different approach or if you’d prefer to make the changes manually. The key changes we need to make are:

2

Hey, have you got a screenshot of this?
For context, the model usually doesn’t know if the apply functionality worked or not!

3

image
image333×864 50.9 KB

I’ve noticed that once the conversation reaches a certain length, this issue pops up every time, forcing me to start a new session and redo everything from scratch.

It’s pretty frustrating and ends up using more tokens just to address the same problem.

4

image
image322×499 25.8 KB

Or, you could keep asking the AI if the edits went through, and while it might say it did, it’s not always reliable.

5

We do have an issue with context missing when a conversation gets too long, so this could be a side effect of that!

I’d recommend routinely starting a new composer session for now, and we will hopefully have a fix for this soon.

6

This continues to be a debilitation problem. Cursor needs to find a way to manage context for longer more complicated conversations when debugging complex code.

7

Hey, whats the issue you are seeing?