Restore checkpoint

[naveenlearner38]

Hi,

This is not an issue. I have been working with Cursor for a long time. Your method is excellent, and it works very well. But most of the time, I use the restore checkpoint. Because the cursor feature might not work well. But in this method, if I restore it, it goes back a few prompts before what works well.

If found solution for restore check point it would be good,

[saitrogen]

hey @naveenlearner38 bro i just tryed this ,but i couldnt get this - can u help me
no python prompt input popup - can u help me to figure this out . am new to this

[perrypixel]

@saitrogen have you made the rule to always? and this works best with claude and gemuni 2.5 pro models

[perrypixel]

@naveenlearner38 using git commit is the only solution i found for this problem. you can change python script to add git commit before asking userinput so a commit is made before you make any changes.

[saitrogen]

@saitrogen have you made the rule to always? and this works best with claude and gemuni 2.5 pro models

Yes bro , did , am using pro free trial. Is that why it is?
I opened new project and it worked there but didn't worked on a working directory

[perrypixel]

@saitrogen If you have rules other than the user input rules, use the .cursorrules file for these rules

[saitrogen]

it some worked bro , i think it happend why i explicitly selected gemini 2.5 pro. thanks
i have a suggestion regarding the hack -

thinks i noted :

• not able to copy/paste prompt into the command terminal (py input) -
• no image uploading

suggestion :

• instead of direct prompt from the command line we should use a markdown file as input (write the prompt we what the python code just read it and feed it to the i ) in this way we can prompt lot more.

• also we can upload image inside that file (assuming in base 64 format or something i dont know how cursor takes the input images)

does that work :) ?

[perrypixel]

Using a markdown file is a great idea. However, uploading images during a conversation isn’t ideal, it's better to upload images only during the initial prompting. Uploading images mid-conversation can cause the model to lose context. I tried this with another tool, and it seemed to forget what it was doing—possibly due to hitting the context limit or something similar

[saitrogen]

I see , so are u working on the markdown 😹😹 let's rock it.

I like to try do that my self - could u tell where should i start? Just ask ai?
Am new to this whole just started using cursor yesterday.

One more thing: check out pocketflow frame work (a 100 line framework for building ai agents with ai agents)
Might come in handy somedays👍🏻👍🏻 )