skills.rest
solitude6060
solitude6060COMMUNITY
Explore all skills by this author →
WebsiteGitHub

triple-review

Orchestrate a three-reviewer GitHub PR review with severity calibration and TDD fix workflow.

Updated May 18, 2026
Search Tags:#code review#pull request#orchestration#triage#GitHub CLI#severity calibration#TDD fix cycle
One-click install
npx skills add https://github.com/solitude6060/Yao-skills --skill triple-review
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: triple-review
Source: https://github.com/solitude6060/Yao-skills/tree/main/skills/triple-review
Command: npx skills add https://github.com/solitude6060/Yao-skills --skill triple-review

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It reduces the chance of shipping defects by running an orchestrator-aware three-reviewer code review with severity calibration and a TDD fix loop.

Core Features & Use Cases

  • Orchestrator-aware reviewer selection: Chooses three external reviewer lanes based on whether Claude Code is orchestrating via codex-family/agy/gemini or Codex is orchestrating via claude/agy/gemini, avoiding self-review unless explicitly requested.
  • Invariant-first prompt structure + diff-correctness: Enforces a consistent review format (verdict, severity-grouped findings, and risk callouts) anchored to project invariants from CLAUDE.md/SPEC, then appends a git diff for concrete, file/line-based critique.
  • Triage and test-driven repair cycle: Triages disagreements using invariant-first heuristics (while requiring factual verification via Read/Grep/Glob), writes failing tests, applies the smallest fix, runs full checks, and archives all review artifacts.

Quick Start

Run: /triple-review <PR#> to perform a three-reviewer PR review, produce severity-triaged findings, and drive a test-first fix cycle.

Frequently Asked Questions about triple-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate a multi-reviewer code review for a GitHub pull request?▼

Automated multi-reviewer code review for a GitHub pull request is orchestrated by assigning three external reviewer lanes to analyze the diff, calibrate severity, and triage disagreements. This process increases defect detection and reduces missed bug classes.

What is orchestrator-aware triage in a pull request review?▼

Orchestrator-aware triage identifies the current orchestrator, such as Claude Code or Codex, and excludes it from the three external reviewer lanes. This prevents self-review and ensures an unbiased, invariant-first critique of the code changes.

How do I run a test-driven fix cycle after triaging PR review findings?▼

Run a test-driven fix cycle by writing a failing test that reproduces the triaged defect, applying the smallest possible fix, and executing full checks. All review and fix artifacts are archived for verification.

Can I use GitHub CLI to format severity-grouped findings for a pull request?▼

Yes, GitHub CLI integrates with the review workflow to pull pull request diffs and enforce an invariant-first prompt structure. It generates verdicts, severity-grouped findings, and risk callouts anchored to project specifications.

Does this three-reviewer PR review work for documentation-only changes?▼

No, the three-reviewer PR review is designed for ready-to-review, non-docs-only pull requests. It requires actual code diffs to perform concrete, file and line-based critique and drive the test-driven repair cycle.

Related Skills

Explore More Software Engineering Skills→
rodrigotoledorodrigotoledo

gh-fix-ci

Identify failing GitHub PR checks and extract failure snippets from Actions logs.

Community
Advanced
WaterplanAIWaterplanAI
31

dr

Forward all arguments from /dr to /dry-run for command validation.

Official
Basic
baphledbaphled
1

check-compliance

Run linting, formatting, security scans, and tests before committing code.

Community
Intermediate
tmalcolm-0607tmalcolm-0607

pr-review

Automate Azure DevOps pull request review with security checklists and risk scoring.

Community
Advanced