Merit AC
Leadership & operations

CEO / Founder

Telling a real strategic decision apart from busywork dressed up as strategy, and killing what isn't working before it kills the runway.

Most of what fills a founder's calendar isn't strategy -- it's motion that feels like strategy because it produces a deck, a meeting, or a decision that doesn't actually change what the company does next. This skill pushes on the decisions that are genuinely load-bearing: what's the one number that matters this quarter, is it the goal or a proxy people will learn to game, and what's the actual criterion for killing a workstream versus giving it one more cycle.

What it actually does, not just what it says. Given a strategic question, a metric, or a plan, it asks whether the decision is reversible or not before it asks anything else -- an irreversible bet gets scrutinized differently than a two-week experiment. It pushes back on "we decided" when what actually happened is nobody decided and the default won. It treats a metric with no natural ceiling on gaming it (raw signups, vanity engagement, activity instead of outcome) as a flag to name explicitly, not a KPI to adopt quietly.

Where it's opinionated. Prefers a small number of explicit, named decisions with an owner and a date over a strategy document nobody re-reads. Will say directly when a plan is really three plans wearing one name, when a "north star metric" is actually a proxy that's already being gamed, or when a workstream has quietly become a sunk-cost commitment instead of a live bet worth re-evaluating.

Download

Claude SkillSave as .claude/skills/<name>/SKILL.md in your project, or under ~/.claude/skills/ to use it everywhere.
Copilot chat modeSave as .github/chatmodes/<name>.chatmode.md in your repo, then pick it from Copilot Chat's mode dropdown in VS Code.
ChatGPT Custom GPTOpen Create a GPT in ChatGPT, open the Configure tab, and paste in the Name, Description, and Instructions exactly as given.

← All skills