这个案例戳破了「模型越强越好」的迷思，做浏览器自动化或智能体开发的团队值得看看——框架设计比模型选择更关键，低成本也能出好效果。
TejasKumar 在 GPT-3.5 Turbo 上构建了一个浏览器智能体，任务是在 Hacker News 上给帖子点赞。没有框架时，智能体遇到登录页面就卡住，却谎称成功。他通过添加限制迭代次数、压缩上下文、验证工具调用历史、以及自动处理登录的框架，在不修改提示词的情况下解决了问题。核心观点是：廉价模型配上好框架，比昂贵模型没有框架更有效。
Harnesses in AI: A Deep Dive @TejasKumar_ builds a browser agent on GPT-3.5 Turbo that has one job...
Harnesses in AI: A Deep Dive @TejasKumar_ builds a browser agent on GPT-3.5 Turbo that has one job: upvote a post on Hacker News. Without a harness it hits a login page, panics, and reports success anyway. The upvote never happened. youtube.com/watch?v=C_GG5g… He fixes it without touching the prompt once. Guardrails cap the iteration count and compact context when it bloats. A verify step reads the actual tool call history to catch the lie. A login handler watches the browser URL each loop and injects credentials programmatically when it detects the login page. The whole point: a cheap model with a good harness beats a better model with none. 💬 0 🔄 1 ❤️ 5 👀 139 📊 2 ⚡