principle-build-the-lever — Claude Code skill: build the tool, not
A skill that pushes the agent to write a codemod, script or generator for non-trivial work instead of editing by hand, so the result can be rerun.
npx skills add backnotprop/pstack --skill principle-build-the-lever -a claude-codeWhat principle-build-the-lever does
The rule is simple: when the work isn't trivial, build the tool that does it. That covers bulk edits, migrations, analyses and checks. Hand-done changes can only be re-verified by redoing them, while a script reruns for free and leaves a reviewer one artifact to read and run again.
Start by doing the first unit by hand to learn the recipe, then build the lever and prove it — rerun it on that unit and diff against your manual version. Edits get a codemod or script, repetitive files get a generator, analysis gets a dump-to-sqlite query, verification gets a rerunnable check. The tool has to be safe to rerun. If it can process every unit in one pass, run it yourself rather than fanning out delegates to hand-apply what a script already does.
When work does go 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. Keep it outside the delegates' write scope, or the contract gets quietly edited. The skill ships with disable-model-invocation: true, so it acts as a rule rather than something the model calls on its own. The bar is triviality, not repetition — a one-off still earns a lever when the lever is what makes the work checkable. It advises the smallest script that does or proves the job, never a framework.
What it helps with
- rewrite identical API calls across a repo
- run a schema migration without missing files
- generate repetitive files instead of copy-pasting them
- leave reviewers an artifact they can rerun
- verify a bulk edit without redoing it by hand
Example prompts for the agent
- Build a codemod for this migration instead of editing files one by one
- Write a script that checks every changed file and show me how to rerun it
- Turn the recipe and verification contract into a skill for the subagents
How to install principle-build-the-lever
1Skills CLI (Vercel)
npx skills add backnotprop/pstack --skill principle-build-the-lever -a claude-code2Manually: put the files in place
~/.claude/skills/principle-build-the-lever/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/principle-build-the-lever/
1Skills CLI (Vercel)
npx skills add backnotprop/pstack --skill principle-build-the-lever -a codex2Manually: put the files in place
.agents/skills/principle-build-the-lever/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add backnotprop/pstack --skill principle-build-the-lever -a cursor2Manually: put the files in place
.cursor/skills/principle-build-the-lever/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install backnotprop/pstack principle-build-the-lever2Manually: put the files in place
.github/skills/principle-build-the-lever/SKILL.md1Skills CLI (Vercel)
npx skills add backnotprop/pstack --skill principle-build-the-lever -a gemini-cli2Platform CLI
gemini skills install https://github.com/backnotprop/pstack --path skills/principle-build-the-leverManually: put the files in place
.devin/skills/principle-build-the-lever/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add backnotprop/pstack --skill principle-build-the-lever -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about principle-build-the-lever
How do I install principle-build-the-lever in Claude?
Run npx skills add backnotprop/pstack --skill principle-build-the-lever -a claude-code. The skill file lands at ~/.claude/skills/principle-build-the-lever/SKILL.md.
What does principle-build-the-lever do?
It makes the agent build a tool for the task: a codemod, script, generator or a skill for subagents. Instead of hand work you get an artifact a reviewer can read and rerun.
When does the skill allow skipping the tool?
Only when the task is trivial — a couple of obvious edits you can see at a glance. Repetition isn't the test: a one-off still earns a lever when the lever is what makes the work checkable.
Source text
SKILL.md · MIT
--- name: principle-build-the-lever 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 isn't trivial, build the tool that does it instead of doing it by hand. A codemod, generator, or script does the work the same way every time and reruns for free. 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". Default to building the lever. Skip it only when the task is trivial, a couple of obvious edits you can see at a glance. - 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. - 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. 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. 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 and reviewability on the work in front of you. For scripting the verification itself, see [Prove It Works](../principle-prove-it-works/SKILL.md).
© backnotprop · MIT · backnotprop/pstack
