troymaclure / LFG

LFG, It Really Whips the Llama's Ass 🦙🦙🦙🦙

LFG

LFG, It Really Whips the Llama's Ass 🦙🦙🦙🦙

LFG is a command-line tool that intelligently helps you find the right terminal commands for your tasks. Such sales pitch. This interface is using GPT-4o as an engine.

Why?

• Firstly, this was created to test Ollama -> Groq
• I do not like the Github Copilot command-line
• Quicker than using Gemini/ChatGPT/Google directly via the browser interface
• Easier to find what needed without opening man pages
• NEW: Changing to GPT-4o model which is free

However, never trust the output entirely.

Installation

# install pipx
brew install pipx

# add pipx binaries to path
pipx ensurepath

# restart your terminal
# install LFG
pipx install lfg-llama

Usage

This executable is using OpenAI, that means you need and API token.

GPT-4o is free to use.

Add the token to your .bashrc/.zshrc and reload your terminal.

OPENAI_API_KEY={replace_me}

$ lfg query

Now you can use the executable

lfg "kill port 3000"

# Kill process listening on port 3000
lsof -i :3000 | xargs kill

Change the LLM

$ lfg "list ec2 pipe json jq get name" -m llama370b

# List EC2 instances with name

aws ec2 describe-instances --query 'Reservations[].Instances[]|{Name:Tags[?Key==`Name`]|[0].Value,I
nstanceId}' --output text | jq '.[] | {"Name", .Name, "InstanceId", .InstanceId}'

This command uses the AWS CLI to describe EC2 instances, and then pipes the output to `jq` to format the output in a JSON-like format, showing the instance name and ID.

Development

pip install --user pipenv
pipenv --python 3.11
pipenv install

pipenv run lfg "kill port 3000"

TODO

• Fix the setup and pyproject file, including github workflow for releasing the package