Skip to content
Agentmart
Submit a tool

principle-build-the-lever

Apply to any non-trivial work, whether one-off or bulk, including edits, migrations, analyses, and checks.

Skill by hellsan631

IndexedMIT
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/hellsan631-principle-build-the-lever

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Apply to any non-trivial work, whether one-off or bulk, including edits, migrations, analyses, and checks. Build a codemod, script, generator, or skill for subagents that does or proves the work. The resulting file gives reviewers something they can rerun.

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/principle-build-the-lever

What's inside

Skill name
principle-build-the-lever
Description
Apply to any non-trivial work, whether one-off or bulk, including edits, migrations, analyses, and checks. Build a codemod, script, generator, or skill for subagents that does or proves the work. The resulting file gives reviewers something they can rerun.
License
MIT
Instructions
23 lines in SKILL.md · view on GitHub

Files (2)

  • SKILL.md
  • openai.yaml · agents/

SKILL.md (first lines)

# Build the lever

For work beyond a couple of obvious edits, build a tool that does the job.

A codemod, generator, or script follows the same recipe every time and reruns without repeating the hand work. It also gives a reviewer one artifact to read and run. Checking hand-done changes requires redoing them. A deterministic script turns "trust me" into "run this."

## Pattern

Build the lever by default. Skip it only when the task is genuinely trivial, meaning a couple of obvious edits you can see at a glance.

- Complete the first unit by hand to learn the recipe, then build the tool. Run it on that unit and diff its output against the hand-done version. Make the lever safe to rerun because a reviewer will rerun it.
- Use a codemod or script for edits, a generator for repetitive files, a dump-to-sqlite query for analysis, and a rerunnable check for verification.
- If one deterministic tool can process every unit in a single pass, run it yourself.
- When work does need subagents, write the lever as a skill they all read. Put the recipe, verification contract, and do-not-touch fences in that one artifact so every delegate follows the same version. This avoids re-explaining the contract in each prompt, where versions can drift. Keep the skill outside the delegates' write scope so they cannot edit the contract.
- Applying this principle produces a file. If you cite it, the diff must contain a codemod, script, generator, or delegate skill.
- Commit the lever when the work outlives the session. The next run can then rerun it instead of redoing the work.

## Limits

The threshold is whether the task is trivial. A one-off still earns a lever when the lever makes the work checkable. Per the [Laziness Protocol](../principle-laziness-protocol/SKILL.md), build the smallest script that does or proves the job. Do not build a framework.

[Encode Lessons in Structure](../principle-encode-lessons-in-structure/SKILL.md) turns a recurring instruction into a durable guardrail. Build the Lever improves throughput by doing the current work and reviewability by leaving a rerunnable file. For a script that performs the verification itself, see [Prove It Works](../principle-prove-it-works/SKILL.md).

From hellsan631/skills, MIT.

More from hellsan631/skills

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Aug 29, 2026 (2 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from hellsan631

Cite this page

principle-build-the-lever skill. Agent Mart. https://agentmart.dev/skills/hellsan631-principle-build-the-lever (last updated 2026-10-24)