This project demonstrates an AI coding agent capable of understanding, interacting with, and modifying Python projects autonomously using natural language prompts.
- Built an AI agent that can inspect project files, read file contents, execute Python scripts, and write or modify files safely within a controlled working directory.
- Integrated with the Google Gemini API to enable reasoning, function planning, and step-by-step execution based on user prompts.
- Created modular functions for core actions:
get_files_info,get_file_content,run_python_file, andwrite_file, each including validation to ensure safe operations. - Implemented an interactive workflow where the agent can automatically detect issues in existing code (e.g., bugs in the calculator) and suggest or apply fixes.
- Enabled verbose mode for detailed tracing of the agent’s reasoning, function calls, and results, highlighting the AI’s decision-making process.
- AI-driven code inspection and debugging
- Safe execution of Python code via programmatic function calls
- Modular, maintainable Python design
- Integration with Google Gemini API for autonomous reasoning and action planning
This project showcases AI capabilities applied to real-world coding tasks.