HN 提问：如何在有 Token 预算的情况下“不手动编写任何代码”？
3 分•作者: mc-0•3 个月前
随着一些企业环境从“不惜一切代价最大化代币使用”转变为为代理开发设定严格的代币预算：
一个人应该如何遵守“不惜一切代价使用人工智能”的管理层/领导指示，甚至更进一步：“你不应该手动编写代码”？
这感觉就像一场俄罗斯轮盘赌。只要你有合适的工具、框架、技能、提示等，能够尽可能地使输出确定化，并且确切地知道自己应该构建什么：你就能得到满足验收标准并解决问题的出色输出。
这一切都没有标准化，具体任务目前也没有可预测的定价（例如，“两个新的 CRUD 端点带测试的价格是 1.43 美元”）。那么，是什么阻止了绩效评估只看你仅凭人工智能能交付什么，而不是看你借助人工智能能交付什么呢？
查看原文
With some corporate environments going from "tokenmaxxing at all costs" to now setting strict token budgets for agentic development:<p>How is one supposed to adhere to a managerial / leadership instruction of "use AI for everything" but even more than that: "you shouldn't be writing code by hand"?<p>It feels like this is a Russian roulette scenario. As long as you have the right harness, framework, skills, prompt, etc. to make the output as deterministic as possible AND know exactly what you're supposed to be building: you get a great output that meets acceptance criteria & solves the problem.<p>None of this is standardized and concrete tasks don't currently have predictable pricing (i.e., "Two new CRUD endpoints with tests are $1.43"). So what's stopping performance reviews from valuing what you can deliver ONLY with AI rather than what you can deliver with AI assistance?