省 token 提示词：让 agent 先列文件再动手
一条可复制的提示词，要求 AI agent 动手前先列出所需文件并说明理由、经批准后再做最小改动，作者称能大幅减少无谓读文件的开销。
Your AI agent doesn't burn most of its tokens on your hardest problems. It burns them reading files and context it never needed. If you're vibe coding with Opus 5.5, GPT-6 Astra, or Grok Bot, steal this prompt: "Before opening any files, list the exact files you need for this task and why. Check the file tree or search to find them, but don't open anything else. If you need more, stop and ask me first. Then, after I approve the list, make the smallest, highest-confidence change that fully solves our problem." AI agents tend to over-read because more context feels safer for them. Yet making them justify every file gets you the same work on far fewer tokens. Scroll past this, and you'll keep paying for files your agent never needed.