Agent-Skills.md

Agent Skills: Critical Path

Process for changes with a large blast radius — files with many callers or repo-p95 complexity, and actions that are destructive, hard to reverse, or visible outside the repo. TRIGGER before editing a file whose enrichment context shows callers ≥ 20 or complexity at the repo p95, and before any delete, force-push, schema drop, process kill, push, pull-request operation, shared-infrastructure change, or third-party upload.

UncategorizedID: heyJordanParker/dotfiles/critical-path

Repository

heyJordanParker/dotfiles
heyJordanParker
203

Install this agent skill to your local

pnpm dlx add-skill https://github.com/heyJordanParker/dotfiles/tree/HEAD/packages/agents/skills/critical-path

Skill Files

Browse the full folder contents for critical-path.

Download Skill

Loading file tree…

packages/agents/skills/critical-path/SKILL.md

Skill Metadata

Name
critical-path
Description
Process for changes with a large blast radius — files with many callers or repo-p95 complexity, and actions that are destructive, hard to reverse, or visible outside the repo. TRIGGER before editing a file whose enrichment context shows callers ≥ 20 or complexity at the repo p95, and before any delete, force-push, schema drop, process kill, push, pull-request operation, shared-infrastructure change, or third-party upload.

Critical Path

  • Mistakes here lose Users and revenue directly.
  • One Process: measure blast radius, get Architect approval for dangerous actions, then verify the dependent capability still holds.

1. Measure the blast radius

Use /trace to measure callers and complexity for what you are changing. Read the nearest Claude.md for the boundary it protects.

Treat high caller count and repo-p95 complexity as Critical Path

A file with at least 20 callers or repo-p95 complexity requires this Skill before editing.

2. Confirm with the Architect before dangerous actions

Approval in one Context never extends to another.

IF the action is destructive, hard to reverse, visible outside the repo, or a third-party upload:

Wait for Architect approval before acting

Destructive means delete, drop, kill, or overwrite. Hard to reverse means force-push, git reset --hard, package removal, or continuous integration change. Visible outside the repo means push, pull request, or shared-infrastructure change.

Never bypass safety checks

Do not use --no-verify or --no-gpg-sign to get past a guard.

Root-cause obstacles instead of clearing them destructively

Do not solve an obstacle with a destructive operation.

IF state is unexpected:

Treat it as another Agent's in-progress work until traced

Unexpected state includes unfamiliar files, branches, and locks.

3. Verify the capability held

Exercise what depends on the change and show the observed output.

Verification is observed behavior

The Skill is not complete until the dependent capability ran and the output is captured.