Hacker News Reader

Comment by ionwake | Hacker News Reader

Sorry if this is a noob question, but are you just pasting file strings inbetween those tags? like the contents of file1.js and file2.js?

View on HN

I do something similar, but "raw" markdown instead + filename, so all my prompts end up like this basically:

    Do blah blah blah while taking blah and blah into account. Here is my current code:

    File `file1.js`:

    ```javascript
    console.log('I am number one!')
    ```

    File `file2.js`:

    ```javascript
    console.log("I am number two :(")
    ```

Not sure if I'm imagining, but when I tried with/without the markdown code blocks, it seems to do better when I used markdown code blocks, so wrote a quick CLI that takes a directory path + prompt and creates something like that automatically for me. Often times I send identical prompts to ChatGPT+DeepThink+Claude, compare the approaches and continue with the one that works best for that particular problem, so having something reusable really saved time for this.

Edit: fuck it, in case people are curious how my little CLI works, I threw it up here: https://github.com/victorb/prompta (beware of bugs and whatnot, I've quite literally hacked this together without much thought)

I actually have a perfect tool called siphon-cli for this. It adds the headers in between files and everything. https://docs.siphon-cli.com/

Yeah, like a lightweight version of my prompta CLI :)

What I end up with, is one .md file that uses variables like "$SRC", "$TESTS" and "$DOCS" inside of it, that gets replaced when you run `prompta output`, and then there is also a JSON file that defines what those variables actually get replaced with.

Bit off-topic, but curious how your repository ends up having 8023 lines of something for concatenating files, while my own CLI sits on 687 lines (500 of those are Rust) but has a lot more functionality :)

Not OP, but practically all of those lines are from a package-lock.json file (6755 lines) and a changelog (541 lines). It looks like the actual source is 179 lines long.

Repomix can take care of this for you. I pack it, cat the file to my clipboard with pbcopy, and just paste it into the prompt.

https://github.com/yamadashy/repomix

I tried the web demo (https://repomix.com/) and it seems to generate unnecessarily complex "packs" for no reason, probably hurts LLM performance too. Why is there "Usage Guidelines" and "File Format" explanations in this, when it's supposed to just be the code "packed"? Better to just have the contents+filename, it'll infer that its directory structure and everything else.

While possibly being strange defaults, both of those are options. Remove the file summary and directory structure, both featured on the UI, and on the CLI tool, and voila, it's in your "better" state. There are also additional compression options beyond those two tweaks.

That's right. I made a VS Code extension to combine all the files I have open into one long string I can copy & paste.

https://marketplace.visualstudio.com/items?itemName=DVYIO.co...

I wrote a script to do exactly this a while ago (with the help of o1 pro). Makes it way easier https://github.com/keizo/ggrab

Reply on news.ycombinator.com