agentleFS
EtanHeyEtanHey/golems/skills/golem-powers/install-runbook-linter/SKILL.md

install-runbook-linter

Skill0 starsChanged 16 days agoApache-2.0View on GitHub

Lint a macOS onboarding runbook for privilege-correctness. Triggers: install runbook, fresh-Mac onboarding, brew cask runbook, admin/Standard user split, happy-camper onboarding.

  • Installs packages

The file

Files
SKILL.md

Install-Runbook Linter

A macOS onboarding runbook is false-green when it reads fine to a human but assigns admin-only steps to a Standard (non-admin) user — the gen16 audit failed exactly here (a Standard user couldn't brew install). This skill turns that audit into a mechanical, replayable gate: runbook_lint.py parses a runbook and reports privilege-correctness violations, with RED→GREEN fixtures (fixtures/, tests/).

Scope

The lint runs before the runbook ships.

Run it

python3 lint_cli.py path/to/onboarding-RUNBOOK.md   # exit 1 if any violation
python3 -m pytest tests/                            # the RED→GREEN gate

Rules

RuleWhat it catches
privileged-in-standard-phaseA privileged command (brew tap/trust/install, --cask, sudo, npm install -g) inside a Standard-tagged phase. Scans only fenced code blocks, so prose prohibitions ("happy-camper must not run brew install") and comments never false-fire. npm install -g is exempt when the runbook establishes a per-user npm prefix (then it writes to user space).
cask-before-prereqsA brew install --cask with no Prerequisites / Phase-0 preflight section earlier in the document.
receipt-not-bundle-verifyEnd-state verification that trusts the cask receipt (brew list --cask) but never checks the on-disk bundle (/Applications/…app, --version, a bundle ls). A receipt is false-green; the on-disk bundle is the truth.
inconsistent-identifier-spellingLoad-bearing identifiers spelled in near-duplicate forms, such as happy-camper vs happy-campep or VoiceBar vs Voicebar. Scans prose and code, but only treats code/code-span identifiers and CamelCase names as load-bearing so heading prose does not false-fire.
prereq-incompletenessDeveloper tools used in fenced code (bun, brew, rustup, node, git, etc.) but not named in a Prerequisites / preflight section.

Phase classification

Phases are split on ## … headers and tagged admin / standard / neutral from role cues (admin, machine-wide vs Standard, happy-camper, user space, without admin). A header naming the Standard user wins (it's the privilege-sensitive one).

Validated against the real artifact

The linter was replayed against the gen17 happy-camper onboarding runbook. The Happy Campr company/launcher name is intentionally distinct from the happy-camper account, while genuine account typos still trip fixtures/red-runbook.md. That delta is the gate: RED on the bad fixture, no noisy findings on the known good patterns, and concrete line numbers for real drift.

Why a linter, not prose

Per the gen-18 doctrine (build the gate, not the prose): a runbook reviewed by eye drifts; a runbook gated by lint_cli.py in CI cannot ship a Standard-user brew install or a receipt-only "verification" again.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.