jimmytoan/ai-coding-survival-guide — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a practical GitHub guide called AI Coding Survival Guide. It should help developers and teams use AI coding assistants without wasting tokens, credits, or review time.

I want it to feel useful and opinionated, not like marketing. Explain the common bad habit of opening a huge repo, picking the strongest model, asking it to fix everything, then spending more prompts cleaning up the mess. Then show a better way, with short docs about context discipline, model selection, agent mode, team guidelines, cost tracking, prompt patterns, alternatives, and good versus bad prompts.

Include reusable prompt templates, a simple benchmark template, and examples people can copy. Keep the writing practical, friendly, and a little funny. The main message should be that AI coding tools are useful, but only when people scope tasks well, choose the right model, ask for a plan first, review diffs carefully, and measure useful outcomes instead of just generating more code.

Want more depth? Deep Reverse