ShellGPT + Phi-3 : OPENSOURCE & LOCAL AI Copilot for your TERMINAL (Github Copilot CLI Alternative)

In this video, I'll be telling you that how you can create your own Github Copilot CLI Clone and Stop Paying for Github's Copilot with this New, Local and Opensource alternative called ShellGPT. We'll be combining it with Microsoft's Phi-3 mini model to make sure that it works even without GPU. It's a really cool alternative to Github's Copilot. This alternative can be used with any opensource LLM, OpenAI models or Claude models such as GPT-4O, Claude-3, CodeQwen, Mixtral 8x22b, Mixtral 8x7b, GPT-4, Grok-1.5, ContinueDev & Gemini Code Assist.
[Key Takeaways]
🎯 Discover Cost-Saving Alternatives: Learn how to bypass GitHub Copilot's $10 subscription fee by using ShellGPT and Phi-3 to make your own Copilot CLI alternative
🔧 Maximize Command Efficiency: Understand the major drawback of GitHub Copilot's terminal bash command feature and explore how to overcome it with Shell GPT's advanced capabilities.
🤖 Meet Shell GPT: Get introduced to Shell GPT, an open-source alternative to GitHub's Copilot CLI, which connects with OpenAI models and local models via Ollama for seamless shell command generation.
💡 Simplify Complex Commands: Find out how Shell GPT can assist with intricate bash commands, ffmpeg scripts, and log filtering using grep, making shell scripting accessible to everyone.
🚀 Easy Installation Guide: Follow a step-by-step tutorial to install Ollama and Shell GPT, and learn how to configure it with the Phi-3 mini model for efficient performance on any computer.
🔍 Interactive Command Execution: Experience Shell GPT's interactive chat feature, allowing you to accept, reject, or refine generated commands, and even execute them directly from your terminal.
📂 Powerful Code Generation: Discover how Shell GPT can create, save, and run code snippets, and generate commit messages from Git diff details, enhancing your coding workflow.

[@SeeFoodDie]

Thank you AICodeKing. Btw I recommend adding the link to the main repo to help beginners easier. This use case has a huge amount of possibilities and since Phi3 is tiny, it can be run with other workloads side by side. Fantastic 🎉

[@alekjwrgnwekfgn]

This relaxes me- the voice.. but won’t robots be doing all this in 6 months?

[@alekjwrgnwekfgn]

What are you doing Dave?

[@1conscience0dimension]

ok I did it with llama3.1:8b it's working.

[@el-tono]

sgpt works fine except this error of litellm in every single response: Model=qwen2:0.5b not found in completion cost map. Setting 'response_cost' to None

[@sillybilly346]

What IDE are you using?? Thanks and great vid!

[@sanapri-uf5vf]

Is that vscode?

[@Mcmarshdzwimbu-ht2md]

helloi am a huge fan , can u please help me i cant find the config file for shell gpt

[@AICodeKing]

It should be at this path: ~/.config/shell_gpt/.sgptrc

[@AnugrahPrahasta]

Your closing music... brain washing.... haha