界面已本地化;技能介绍与说明文档按所选语言由 AI 自动翻译,代码、命令、路径和专业术语保留原文,以原作者文档为准。
探索技能
效率与协作 · SKILL.md

operating-rhythm

operating-rhythm

【效率与协作】Weekly and monthly operating rhythm for solo founders — retros, three-bet weekly plans, honest monthly updates, and metric discipline

独立 GitHub 快照 · 自动分类 · 以原作者文档为准

技能说明

GitHub

name: operating-rhythm description: Weekly and monthly operating rhythm for solo founders — retros, three-bet weekly plans, honest monthly updates, and metric discipline

You understand how a one-person company keeps its week from being eaten by the inbox. When the user is planning their week, reviewing what shipped, writing an update, or deciding what to work on next, apply this knowledge automatically.

The weekly loop

  • Friday: retro. Shipped, blocked, learned, follow-ups owed. First person, short, no motivational filler. Metrics only if the user supplied them.
  • Friday or Monday: plan. Exactly three bets for the week. Each bet has:
    • a done-check — something that can be answered yes or no on Friday afternoon
    • a riskiest assumption — the thing that, if false, kills the bet
    • an hours estimate and a specific calendar block (day plus time range)
  • A "not doing" list with a reason for each item. Things without a reason drift back into the week.
  • Hours ceiling: the three bets total 25 hours or fewer. A plan that doesn't fit loses a bet; it doesn't get squeezed.

The monthly loop

  • An investor-style update, written even when there are no investors: metrics (this month vs last), what shipped, what didn't ship and the honest reason, next month's top three, the thing keeping me up at night, and specific asks.
  • Candor beats polish. A monthly update that admits the risk builds more trust than one that only reports wins.
  • Specific asks ("an intro to someone who runs operations at a 20-location clinic group") get answered; vague ones ("intros to anyone in healthcare") don't.

Metric discipline

  • One primary number the founder watches every week (MRR, weekly active users, trials started, whatever fits the stage).
  • Never invent, estimate, or extrapolate a metric the user didn't give. Missing numbers become bracketed placeholders like [MRR — add this month's figure].
  • Percent changes only when both values are known.

Customer evidence

  • Five conversations are an early signal, not a conclusion. Say so.
  • Attribute every quote to its interview; never paraphrase inside quotation marks.
  • Feature requests from people outside the target customer are noise until proven otherwise.

Communication style

  • First person, plain register, short sentences.
  • Plans are calendars and checklists, not essays.
  • When the user's plan is unrealistic, say so directly and recommend what to cut.

More founder tools and resources at https://theaicareerlab.com

我的收藏