principle-build-the-lever ยท diff
git:20260527.b3eabf1 to git:20260819.4b32d9d
12 added, 9 removed. Audit A to A.
---
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."
+ description: "Apply to any non-trivial work, not just bulk work: edits, migrations, analyses, checks. Build the tool that does it or proves it (codemod, script, generator, or a skill your subagents follow) instead of working by hand. The tool is the artifact a reviewer can rerun."
disable-model-invocation: true
---
# Build the Lever
- When the work repeats, build the tool that does it instead of grinding by hand.
+ When the work isn't trivial, build the tool that does it instead of doing it 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.
+ **Why:** Two payoffs. Throughput: a codemod, generator, or script does the work the same way every time and reruns for free. Confidence: the tool is one artifact a reviewer can read and rerun to check the work. Hand-done changes can only be re-verified by redoing them. A deterministic script turns "trust me" into "run this".
- **Pattern:** When you would otherwise repeat a transform, a check, or a setup more than a handful of times, build the lever instead.
+ **Pattern:** Default to building the lever. Skip it only when the task is genuinely trivial, a couple of obvious edits you can see at a glance.
- - Do the first few by hand to learn the exact recipe, then build the tool. Don't build on a guess.
- - Codemod or script for bulk edits, generator for repetitive files, a dump-to-sqlite query for repeated analysis, a rerunnable check for repeated verification.
- - Commit it when the work outlives the session, so the next run reruns it instead of redoing it.
+ - Do the first unit by hand to learn the recipe, then build the tool. Prove it by rerunning it on that unit and diffing against your hand-done version. Make the lever safe to rerun. A reviewer will.
+ - Codemod or script for edits, generator for repetitive files, a dump-to-sqlite query for analysis, a rerunnable check for verification.
+ - A deterministic lever beats fan-out. If the tool can process every unit in one pass, run it yourself; don't fan out delegates to hand-apply what a script can do.
+ - When you fan work out to subagents, write the lever as a skill they all read: the recipe, the verification contract, and the do-not-touch fences in one artifact, so every delegate inherits the same hardened version instead of re-explaining it per prompt and watching each one drift. Keep it outside the delegates' write scope so they can't quietly edit the contract.
+ - Applying this principle produces a file. If you cited it and there is no codemod, script, generator, or delegate skill in the diff, you didn't apply it.
+ - Commit the lever when the work outlives the session, so the next run reruns it instead of redoing it.
- **Balance:** Leverage, not gold-plating. The [Laziness Protocol](../principle-laziness-protocol/SKILL.md) still holds. Build the lever only when it pays for itself across the remaining work, never for a one-off.
+ **Balance:** The bar is triviality, not repetition. A one-off still earns a lever when the lever is what makes the work checkable. Per the [Laziness Protocol](../principle-laziness-protocol/SKILL.md), build the smallest script that does or proves the job, never a framework.
- Distinct from [Encode Lessons in Structure](../principle-encode-lessons-in-structure/SKILL.md), which makes a recurring instruction a durable guardrail. This is throughput on the work in front of you.
+ Distinct from [Encode Lessons in Structure](../principle-encode-lessons-in-structure/SKILL.md), which makes a recurring instruction a durable guardrail. This is throughput and reviewability on the work in front of you. For scripting the verification itself, see [Prove It Works](../principle-prove-it-works/SKILL.md).