Command-Line AI Agent with Tool Integration

This project provides a command-line chat agent that integrates with a language model and supports tool-based interactions for file operations and calculations. Users can interact using natural language or explicit slash commands, with tab completion enhancing usability and efficiency.

Installation

Clone the repository and install dependencies:

$ pip install -r requirements.txt

Set your API keys depending on the provider:

$ export GROQ_API_KEY=your_key_here
$ export OPENROUTER_API_KEY=your_key_here

Usage

You need a sentence introducing every code block (### titles not good form); you also need to include the $ in front of every shell command; (it is sometimes acceptable to not include the $ if every code block on a page is only a shell command with no output, but that is not the case here, so you need the $ on your commands)

$ chat
chat> what files are in the .github folder?
The only file in this folder is the workflows subfolder.

You can pass a message directly:

$ chat "what files are in the .github folder?"
The only file in this folder is the workflows subfolder.
...

Specify which model provider to use:

$ chat --provider openai
chat> What model are you?
I am GPT5.2 provided by OpenAI

Supported providers:

• groq (default)
• openai
• anthropic
• google

Debug Mode

Debug mode prints tool usage whenever a tool is invoked.

$ chat --debug
chat> /ls .github  
[tool] /ls .github  
The only file in this folder is the workflows subfolder.

Example Queries on Projects

You should never have a section without a sentence in it; the examples above are also usage examples, so this section needed a more descriptive title

Markdown Compiler

This example demonstrates how the chat tool can analyze a codebase by searching for specific patterns across files.

cd test_projects/Markdown-to-HTML-compiler
chat
chat> does this project use regular expressions?
No. I grepped the project files and did not find any use of the `re` library.

This example is useful because it demonstrates how the agent uses the grep tool to analyze code structure across files.

Ebay Scraper

This example demonstrates how the agent can summarize a project and answer higher-level questions about its purpose and implications.

cd test_projects/Ebay_webscrapping
chat
chat> tell me about this project
The project is designed to scrape product information from eBay listings.

chat> is this legal?
In general, scraping public webpages is often legal, although using an official API is usually more reliable and efficient.

This example is useful because it shows the agent can summarize a project and reason about broader implications.

Personal Website

This example demonstrates how the tool can interpret and summarize the contents of a non-Python project.

cd test_projects/abedoya-norena.github.io
chat
chat> what does this project contain?
This project contains the files for a personal website, including HTML and related assets.

This example is useful because it demonstrates that the agent can interpret non-Python projects using file inspection.