// Apply when work is repetitive or bulk: many similar edits, a check you'll rerun, a population to transform. Build the tool that amortizes it (codemod, script, generator) once you know the recipe, instead of grinding by hand.
| name | principle-build-the-lever |
| description | Apply when work is repetitive or bulk: many similar edits, a check you'll rerun, a population to transform. Build the tool that amortizes it (codemod, script, generator) once you know the recipe, instead of grinding by hand. |
| disable-model-invocation | true |
When the work repeats, build the tool that does it instead of grinding by hand.
Why: Doing the same edit a hundred times is slow and drifts into inconsistent mistakes. A codemod, generator, or script does it once, the same way every time, reruns for free, and gives a reviewer one artifact to check.
Pattern: When you would otherwise repeat a transform, a check, or a setup more than a handful of times, build the lever instead.
Balance: Leverage, not gold-plating. The Laziness Protocol still holds. Build the lever only when it pays for itself across the remaining work, never for a one-off.
Distinct from Encode Lessons in Structure, which makes a recurring instruction a durable guardrail. This is throughput on the work in front of you.
Use for "how does X work", code walkthroughs before changing something, and placement / ownership / layering questions ("where should this live", "which package owns this", "is this the right layer"). Explains subsystem architecture, runtime flow, onboarding mental models. Can critique architecture. Use why for motivation.
Use for "interrogate", "adversarial review", "multi-model review", "challenge this", "stress test this code", "find blind spots", or "tear this apart". Four LLM reviewers challenge changes from independent angles.
poteto's agent style for concise, detailed responses, deliberate subagents, unslopped prose, simple code, and verified work. Use for poteto, /poteto-mode, or requests to work in this style.
Spawn three parallel review subagents over the active transcript, surface learnings, and route each to a concrete edit on an existing skill. Use when the user says reflect.
Use for 'why does X work this way', 'why we picked Y', design rationale, regressions, postmortems, or data-backed thresholds. Discovers available MCPs and queries each evidence category (source control, issue tracker, long-form docs, real-time chat, infrastructure observability, error tracking, product analytics warehouse) in parallel, then returns a cited read on decisions and tradeoffs. Use how for runtime behavior.
