Agentic AI

General-purpose Python coding agent with the ability to read, write and run files. Utilizes and maintains internal reasoning (chain of thought) across messages, but it can be also turned off to spare tokens on easier tasks. Session logs are automatically saved in a list of JSON objects, capturing every thought, tool call, and result.

Warning: Although there are path-traversal protection and subprocess timeouts in place to ensure the agent remains within the permitted workspace, apart from that, the agent can do anything on your machine.

🚀 Quick start

1. Clone the repository

git clone github.com/nonlinear-vibes/agentic-AI cd agentic-AI

2. Set up your environment

Create a .env file in the root directory and add your Gemini API key:

echo "GEMINI_API_KEY=your_key_here" > .env

3. Install dependencies

pip install -r requirements.txt

4. Run the Agent

python main.py

🛠️ Project structure

. ├─ functions │ ├─ get_file_content.py │ ├─ get_files_info.py | ├─ run_python_file.py | └─ write_file.py ├─ logs | └─ [saved session logs] ├─ workspace | └─ [your project folder] ├─ call_function.py ├─ config.py ├─ main.py └─ prompts.py

Upon a user reuest, the agent can decide either to generate a response or call for function execution. Each funtion execution's result is returned to the agent and it can decide again which action to take, and so on in a loop. Once it decides to respond with a text, the user can prompt it again.

⚙️ Configuration

The agent's behavior can be toggled in config.py:

• MAX_CHARS - Maximum number of characters that can be read from a file in a single read function call.
• WORKING_DIR - Name of your working directory. Strict path verification ensures that the agent cannot operate outside of this directory.
• MAX_ITERS - Maximum number of function call iterations in a single response.
• VERBOSE - If set to True, function calls and responses are printed to the console.
• THINKING - Enables thinking capability.
• KEEP_THOUGHTS - Keeps the agent's internal thought process in the session history, enabling better chain of thought context on the expense of higher token traffic.
• THINKING_TOKEN_LIMIT - How many tokens can be used for thinking in a single round of messages.

🔧 Agentic functions

The agent currently has the following capabilities out of the box:

• File Exploration: List files and directories with metadata.

• Surgical Read: Read specific line ranges of files. Surgical File Access: Advanced file-reading tools allow the agent to read specific line ranges, enabling it to handle large codebases efficiently.

• File Writing: Create or overwrite files with automatic directory creation.

• Code Execution: Run Python scripts and capture STDOUT/STDERR/Tracebacks for self-debugging.