- SignalDesk1 hr ago
Original Summary
I posit that:<p>1. I am faster using agentic coding than without<p>2. I am 10x faster using many agents in parallel<p>3. When I hit a limit (5h, week), its a good time to take a break - and focus on some slower work or step away.<p>4. If I can reliably produce quality output with agents, then I should attempt to automate the workflow from idea to QA.<p>5. If I can automate how I work, I can sleep while I work.<p>6. If I can sleep when I work, I can get more done, while working less.<p>7. If I can token max while retaining and continually improving the process and output quality, I can work less and earn more.
中文概览
中文标题: Ask HN:你会把 token 用量拉满吗?为什么?
发帖者提出一系列观点:使用代理式编码比不用更快;并行使用多个代理可快 10 倍;触及 5 小时或每周限额时适合休息或做较慢工作;若能稳定产出高质量结果,就应尝试自动化从想法到 QA 的流程;若自动化工作方式,就能在睡觉时也工作;若能在工作时睡觉,就能做得更多、工作更少;若能在保持并持续改进流程和输出质量的同时把 token 用量拉满,就能少工作多赚钱。
- 情报分类:技术学习与提效
- 分类依据:讨论代理式编码、多代理并行与自动化工作流的提效
- 信息来源:Hacker News 新项目
- 发布时间:2026/9/25 09:22:13
- No replies yet