Stats
Actions
Tags
From arclight
Independent code review pass. Dispatch a reviewer subagent against the change set. Reviewer checks bugs, security, performance, idioms, test coverage, and spec adherence. Returns issues categorized by severity. You then triage and address findings before ship.
How this skill is triggered — by the user, by Claude, or both
Slash command
/arclight:reviewThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
External review pass. Same author can't catch their own blind spots.
External review pass. Same author can't catch their own blind spots.
status=pass)If any fail, return status=blocked with the gap.
git diff <base>..<head>. Record the file list and total LOC.reviewer-prompt.md. Returns issues by severity (must-fix, should-fix, nice-to-have) + verdict. Both prompts now include a Contract Verification category that walks every spec claim and verifies the impl honors it.open-pass-prompt.md. Looser prompt, higher recall. Catches what the structured prompt misses (out-of-scope edge cases, platform issues, spec gaps, contract violations).must-fixshould-fixnice-to-have
The merged verdict is needs-changes if any merged item is must-fix, else ready-to-ship.must-fix — blocker, do before shipshould-fix — fix in this PR if possible, defer with a ticket if notnice-to-have — log to follow-ups, do not blockverify after fixes.Reviewer items marked "advisory" default to nice-to-have unless the user disagrees.
You MUST refuse to advance to ship if:
pass{
"issues": [
{ "severity": "must-fix" | "should-fix" | "nice-to-have", "where": "<file:line>", "what": "<>", "fix": "<>" | null }
],
"addressed": <int>,
"deferred": [{ "issue": "<>", "ticket": "<>" }],
"ready_to_ship": true | false
}
Return to the orchestrator. If ready_to_ship=true, the orchestrator advances to ship. Otherwise, return to implement (or design if the issue exposes a spec gap).
npx claudepluginhub nurseapp/arclight --plugin arclightGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.