I find Claude Chrome's extension to be much more powerful than Cursor's current browser integration. Especially if I need to debug something that requires being logged in to a system that uses SSO for example. This repository will allow you to use Claude's Chrome extension from Cursor against the currently selected tab in your already-open default browser. I created this so that I could take advantage of Claude's Chrome extension from Cursor: Cursor calls a local helper script, the script activates the existing default Chromium browser session, opens the Claude side panel, sends a prompt, reads Claude's final response through macOS Accessibility, and returns that response to Cursor so Cursor can continue coding or editing files.
Using whisper.cpp's stream build I used a few keywords to decide when to send what data to chatGPT which is then written back to your terminal. Crazy with the $150 per Interview web application solutions I saw in the past.
It's as simple to use as running a build and run command.
After running a command in the terminal, I often find myself needing to run the same command again, without wanting or being able to easily re-type the command or use the arrow keys (think mobile). This post contains ZSH scripting that keeps track of the last non-empty ZSH command prompt/buffer and runs it whenever I send a few empty lines by pressing the enter key 2 times in a row.
Yes, I know we should all be good enough to not need to ever use console.log statements, or to even iterate for that matter but sometimes I find myself tired and working on a project and just want to double check my data structure is in the correct state. If I use console.log I risk the chance of forgetting that line of code and submitting it to LeetCode which will cause a failure. Using this comment based approach makes logging easier and is "LeetCode safe" in that no console.log statements will ever be executed.
In this blog post, I show how I created a preprocessor that takes any line starting with “// log -” and creates a temporary intermediary file where those lines are replaced with console.log statements to print any expressions following “// log -”. This approach is useful in several ways. It’s easier than using console.log directly, and when run without using the preprocessor, these log lines are simply ignored. The entire setup is straightforward, utilizing a Node.js environment variable/option to run the preprocessor.
Previously, I used a custom bash script to compile repository files into a single context file, but Zed AI’s implementation is far superior, offering a more streamlined and effective solution. But it never hurts to have an alternative method laying around. Like this little shell script..
It opened my eyes to a new style of game development and a new way to think about dependency injection and types.
I have always thought that a majority of the hardest software engineering problems relate in some way to gaming and when I see solutions like this it makes me think I am right.
I made a little proof of concept that used whisper cpp and ChatGPT to take “command” requests and generate Apple scripts which I could then run in OS X
It actually works somewhat well. I think with some more work and thought, something like this, could actually be useful.
I had my phone stolen and basically did the same thing. I didn't want to approach anyone for fear of starting some altercation. Wish there was a better way to go about this.
With my phone, if I had another device maybe I could have made it make a sound and then started a conversation with the person that way? I didn't know who exactly had it, although I am pretty sure I saw the person just walk away with it. Wasn't sure enough to make an accusation though.
I tried to implement this using OpenAI's embeddings then using cosign similarity between produces vectors instead of the vector DB ( as shown in OpenAI's example code here ). I do the same thing where I take the highest ranking code snippets (vectors) and include them in a prompt to ChatGPT with an original prompt. My code is a mess since I just hacked it together after a long day of work but its a small file. Like 100 lines.
I used TreeSitter which I thought was pretty awesome though because it allows for parsing a TON of different languages. I had to parse the languages to create the different code snippet strings. I don't want to create a code snippet of half a function for example..
So TreeSitter parses the code into an AST and I send each different AST node to OpenAI to get the vector (I optimized this so multiple nodes of the same AST type are combined). Send the prompt to OpenAI to get a vector. Find the most similar code snippets to the prompt and include them at the top of a prompt to ChatGPT.
From what I understand this seems useful if you have a model that will accept a large or unlimited number of tokens. I was looking into doing the same thing with ChatGPT and went with ada to find snippets related to the prompt and then to include those with a prompt to ChatGPT: https://bbarrows.com/posts/using-embeddings-ada-and-chatgpt-...
Using OpenAI's brand new API's to search through any of your local codebases (that ChatGPT is not trained on!). This is what Github CoPilot View is working on as well actually: https://githubnext.com/projects/copilot-view/
Using FFMPEG to record a short audio clip, then using the openai CLI this post shows how to transcribe your audio and send it to ChatGPT using a short script.
Quick script to record audio and send it off to OpenAI for transcription and then to ChatGPT
This script records 10 seconds of audio using FFMPEG and then uses the openai cli tool for the rest.