🎉 The Instagram Growth Blueprint: How to Gain 1,000 Real Followers in 30 Days Without Paid Ads

gittech. site

for different kinds of informations and explorations.

AI Creating AI: Open-Sourcing 'Ask' – Your Agentic Command-Line Assistant

Published at
Jan 16, 2025
Main Article
https://github.com/sfarrell5123/ask

ask - Command-Line Assistant

ask is a powerful command-line assistant that was originally created by an AI coding agent, demonstrating agentic capabilities—planning, writing code, testing, and debugging autonomously. It’s designed to streamline terminal operations using Large Language Models (LLMs) and can even write and execute scripts on your behalf.

Read more about the AI origins and screenshots in my LinkedIn post:
This is my AI software agent, that makes software!
or check out the full article:
AI: Making My ‘ask’ Command Line Helper Dude

Features

  • AI-Created & Agentic: Originally generated by a coding agent that autonomously plans, writes code, tests, and debugs.
  • File Operations: Easily write, replace, or insert content in files with intelligent assistance.
  • Command Execution: Execute shell commands with context-aware support and automatic validation.
  • Task Management: Create and execute step-by-step plans for complex tasks, ensuring accuracy and efficiency.
  • LLM Integration: Leverage advanced LLMs like Groq and Deepseek for intelligent decision-making and assistance.
  • Self-Testing & Correction: Automatically checks its own work and corrects mistakes.
  • Data Access: Pipe data in, provide a file reference, or just talk about a file—it will read and interpret it.
  • Re-entrant/Agentic: it will keep calling itself, and correcting, until it gets done
  • Uses AI to help summarise history: which sort of works

UnFeatures

  • sometimes you have to tell it a few times to get on with the plan, it loves to plan
  • sometimes it writes plans for really small jobs, it loves to plan
  • sometimes it emits text, instead of doing "use your tools" helps sometimes
  • "discuss only" helps - if you don't want things trashed
  • if you look away too long, it will init --skynet if you aren't careful. Seriously though , it could trash things, be careful, have backups.

Bad Stuff in the code

  • the history file is horendous
  • needs a --clear-history flag or something (I just use rm ~/.cache/ask/history.json)
  • it's using deepseek currently, smart and cheap, but lets the chinese govt spy on you, seriously - I think this happens
  • probably need openai settings for safety, which given deepseek uses openai api, just change the URL and API key and model, and you're good to go. This would rock on O1.

Installation

  1. Clone the repository:
    git clone https://github.com/sfarrell5123/ask.git
  2. Navigate to the project directory:
    cd ask
  3. Install required dependencies:
    pip install -r requirements.txt
or say pip3.11 install -r requirements.txt
  4. Set up environment variables by creating a ~/.env file in your home directory with the following content:
    GROQ_API_KEY=your_groq_api_key
DEEPSEEK_API_KEY=your_deepseek_api_key
  5. Make the script executable:
    head ask (you might need to change which python you use)
chmod +x ask
cp ask /usr/local/bin/ask

Usage

Simply type your question or command, and ask will handle the rest:

ask "how to list and test the contents of a tar file ?"

Examples

  • File Operations:
    ask "Write 'Hello, World!' to hello.py"
  • Command Execution:
    ask "List all files in the current directory, tell me which is the biggest"
  • Task Management:
    ask "Here is my plan (prd.txt). Read it, create the code, then test it."

Detailed Examples

File Operations

  • Append to a File: 

    ask "explain this file" doSomething.py

  • Read a Specific Line from a File: 

    ask "does this file discuss economics?" countries.txt

  • Delete a Line from a File: 

    ls -lh | ask "which is the most recent py files that I've been working on?"

Command Execution

  • Execute a Command with Output Redirection: 

    ask "list all snapper config, and check the latest backups for each, write me a cron task for it daily"

  • Execute a Command with Piping: 

    ask "update dostuff.py - it's not working "

Task Management

  • Create a Plan with Conditional Steps: 

    ask "Create a plan to deploy the application: 1. Check if the environment is set up, 2. If yes, deploy the application, 3. If no, set up the environment first"

  • Execute a Plan with Validation: 

    ask "Execute the plan to deploy the application and validate each step"

Troubleshooting

  • Common Issues:

    • Make sure the ~/.env file is correctly set up with the required API keys.
    • Place ask in a directory that’s in your PATH (e.g., /usr/local/bin).
    • Ensure the script has executable permissions.
    • Check the terminal output for any error messages and follow suggested fixes.

  • Debugging:

    • Use the --debug flag to enable detailed logging: 
      ask --debug "Your command here"

    Here’s a section you can add to your README to encourage users to share their experiences, inputs, outputs, and suggestions for improving ask:

Community Contributions: Share Your Uses and Ideas 💡

We’re building ask together, and I’d love to see the creative ways you’re using it! If you’ve found an interesting use case or have ideas to make ask even better, here’s how you can contribute:

Share Your Use Cases

Have you done something cool with ask?
Whether it's a unique input/output combo or an unexpected automation win, I want to hear about it! Please share:

  1. Your Input: What did you ask ask to do?
    Example: 

    ask "can you create a bash script to monitor disk usage and send an alert if it exceeds 80%?"

  2. The Output: How did ask respond?
    Example: A generated bash script or the outcome of a complex command.

  3. Your Feedback: How did it work? Was the result what you expected? Any tweaks or improvements you made?

Submit your examples by opening an issue in the repository .

Suggest Ideas for Improvement

Have ideas to make ask even better? Here are some suggestions to spark your creativity:

  • New Features: What’s missing that would make ask more useful to you?
    Example: “Add support for Docker commands” or “Integrate with Kubernetes.”

  • Enhancements: Found a limitation or something that could be smarter?
    Example: “Improve file size handling when passing large files for LLM processing.”

  • Edge Cases: Have you found a scenario where ask struggled or failed?
    Example: “ask didn’t correctly identify my Python script’s error when running a large loop.”

Submit your ideas via GitHub Issues, or start a discussion in the repository.

Why Contribute?

  • Shape the Future: Your input directly influences ask’s growth and capabilities.
  • Learn & Share: Collaborate with others in the community to explore new use cases and ideas.
  • Get Acknowledged: Exceptional contributions will be highlighted in future updates!

I’m excited to see what you create and how we can make ask even better together. Let’s keep pushing the boundaries of what an AI-powered command-line assistant can do!

Frequently Asked Questions (FAQ)

  1. What LLMs are supported?
    ask supports Groq and Deepseek LLMs. You can configure which model to use in the .env file.

  2. Can I use ask for scripting?
    Yes, ask can be integrated into shell scripts to automate complex tasks. It can create new files/scripts in bash or Python, then execute them—or perform a series of commands.

  3. How do I update ask?
    Pull the latest changes from the repository and reinstall the dependencies:

    git pull origin main
pip install -r requirements.txt

Advanced Usage

Below are some real-world terminal scenarios.

  • Linux ZFS Example:

    zpool import | ask "I don't want the data pool to import automatically, but when I do, I want it to import with a root of /mnt/data. Show me how."

  • Install Linux Software:

    ask "Recommend a good, easy-to-manage FTP server for my local network"
ask "That sounds good. Make it so number 1, engage."

  • Coding:

    ask "Take a look at this code and extract all YouTube video URLs from the playlist" ./bulk_transcribe_youtube_videos_from_playlist3.py

  • Image Operations:

    ask "create a 1400x1400 version of maxine.png called maxine_square.png, use your tools"

Contributing

Contributions are welcome! Follow these steps to get involved:

  1. Fork the repository.
  2. Create a new branch for your feature or bugfix.
  3. Commit your changes.
  4. Push your branch and submit a pull request.
  5. I’m keen for collaborators—feel free to help maintain the project!

License

This project is licensed under the MIT License.

Open Source Contribution Guidelines

We kindly ask that if you fork or use this code, you acknowledge the original creator and the AI origins of this tool. You can do so by including a simple note in your documentation or README:

This project is based on the original work by Scott Farrell, which was created by an AI coding agent. 
You can find the original repository at https://github.com/sfarrell5123/ask.

By contributing to this project, you agree to abide by these guidelines and ensure that the original author is properly credited.

Contact

This entire project, known as ask, is a testament to what AI software agents can build—from the first line of code to fully autonomous planning, coding, testing, and debugging. Check out the LinkedIn references above to learn more about its creation journey!

Happy coding with your new AI-powered CLI companion!