principle-build-the-lever

An agent skill by cursor, from cursor/plugins. Tags: analysis, automation, developer-tools, setup, workflow.

What it does

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.

Install

With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:

npx skills add cursor/plugins --skill principle-build-the-lever

Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):

git clone --depth 1 https://github.com/cursor/plugins
cp -r plugins/pstack/skills/principle-build-the-lever ~/.claude/skills/principle-build-the-lever

Safety box score

Grade A, minimal risk: 7 of 100 risk points, rated 2026-10-19. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.

This skill is a procedural guide for building a deterministic “lever” — usually a script, codemod, or generator — to do non-trivial work and rerun it for verification. The shipped file is plain markdown only; the main effect is to encourage local code execution and workspace edits, with no network, secrets, privilege, or opaque components included.

Category levels, from 0 (none) to 3 (high)
Category	Level	Why
Code & shell execution	1 (low)	The skill explicitly tells the agent to build and rerun a "codemod, script, generator" and to "run it yourself," so its normal use may involve executing newly created code, even though no executable ships in the repo.
Network access	0 (none)	There are no download/upload steps, remote URLs, or API calls in the skill text.
Filesystem reach	1 (low)	It expects the agent to produce a file artifact and says "Applying this principle produces a file" and "Commit the lever," which implies edits in the workspace but not outside it.
Secrets & credentials	0 (none)	The skill does not ask for tokens, passwords, dotfiles, keychains, or other credentials.
Privilege & persistence	0 (none)	There is no mention of sudo, global installs, services, shell startup changes, or other privilege-bearing actions.
Instruction hijack surface	0 (none)	The text contains no instruction to ignore higher-priority rules, hide actions, or follow fetched instructions; it is purely procedural.
Irreversible actions	0 (none)	It changes how work is done, but it does not itself send messages, deploy, delete, pay, book, or otherwise act on external systems.
Transparency	0 (none)	The repository content is a single readable markdown file; there are no binaries, encoded blobs, or unreviewable installers.

Before installing

Confirm you are comfortable with an agent generating and running scripts in your workspace.
Check that the repo ships only `SKILL.md` and no native executables or installers.
If you will use it on real code, require a reviewable dry-run or diff from the generated lever before it is applied.

Rated 2026-10-19 by an automated review of 1 file from the skill's repository.

Source

Repository: cursor/plugins (all skills from this repository)
Path: pstack/skills/principle-build-the-lever/SKILL.md
Branch: main
Collection: cursor
Updated: 2026-09-23

automate-me — Use for "automate me", "create/update/refresh my -mode skill", "turn/capture my preferences or working style into a skill".
cli-for-agents — Designs or reviews CLIs so coding agents can run them reliably: non-interactive flags, layered --help with examples, stdin/pipelines, fast actionable errors.
create-verification-skill — Generate a project-local verification skill that drives your app the way a user does — any language, framework, or platform.
setup-benny — Configure Benny and prepare its triage and repro automations.
setup-pstack — Configure which models pstack uses per role and at what reasoning budget.
swarm — Fan out N parallel workers, drain them, and return one report. Use for /swarm, 'swarm this', or parallel coverage, races, gauntlets, and exploration.