Autonomous Agent

This project implements an autonomous agent that can perform tasks based on instructions provided in INPUT.md. It uses the Gemini API to process tasks, manage files, and interact with the environment.

How it Works

The agent operates in a loop:

1. Reads tasks from TASKS.md.
2. Processes each task using the Gemini API.
3. Updates TASKS.md to reflect the completion status.
4. If there are TODOs in the code, it creates new tasks in TASKS.md to address them.
5. Once all tasks are complete and there are no TODOs, it generates a wiki.

Usage

1. Set the GEMINI_API_KEY environment variable.
2. Create an INPUT.md file with a list of tasks.
3. Run the agent: go run main.go [working_directory] (optional working directory).

Files

• INPUT.md: Contains the initial list of tasks.
• TASKS.md: Contains the current list of tasks with completion status.
• main.go: The main entry point of the application.
• agent.go: Contains the agent's core logic.
• tools.go: Defines the available tools for the agent.
• wiki.go: Generates the project wiki.

License

MIT