This repository provides a ruleset and a Python script to enable iterative, multi-turn interactions with AI coding assistants that support tool calls, specifically within agent modes. It aims to maximize the utility of monthly AI tool call allowances by allowing users to chain multiple follow-up commands within a single AI request, thereby achieving more work per allocated request.

How It Works

The core mechanism involves a Python script (userinput.py) that prompts the user for follow-up instructions after the AI completes an initial task. These instructions are then fed back into the AI's context, allowing it to continue working on the task iteratively. This loop continues until the user manually stops the session or the tool call limit is reached, effectively consolidating multiple distinct commands into a single initial AI request.

Quick Start & Requirements

• Copy userinput.py to your project's root directory.
• Copy the contents of rules.md into your IDE's rule configuration (e.g., .cursorrules in Cursor, set to "always").
• Requires an AI coding assistant that supports agent mode and tool calls (e.g., Cursor IDE).
• Designed for tool call-based quotas, not token-based pricing.

Highlighted Details

• Enables up to 25 follow-up commands within a single AI request's tool call allowance.
• Maximizes work done per monthly AI request quota.
• Simple setup via a Python script and IDE rules file.

Maintenance & Community

• Development is supported via ko-fi: https://ko-fi.com/perrypixel.

Licensing & Compatibility

• The repository does not explicitly state a license. Users should assume all rights reserved or contact the author for clarification.
• Compatibility is limited to AI coding assistants supporting agent mode and tool calls.

Limitations & Caveats

Currently, only plain text input is supported; image uploads or file drops are not yet implemented but are planned for future versions. This setup is specifically designed to leverage bundled tool calls and is not intended for minimizing token usage.