Skill Claude CodeCodex
Use this skill for reviewing code changes, diffs, pull requests, completed implementation tasks, or merge readiness when the review should run as an adversarial, self-terminating convergence loop. Reviewer lanes attack the diff, an independent verifier tries to refute every blocker against the real code before it can…
Skill Claude CodeCodex
Compact the current conversation into a persistent handoff document so a fresh agent can continue the work without re-understanding the world.
Skill Claude CodeCodex
An AI development harness is a set of hooks, checks, feedback loops, and rules that guide and verify an AI coding agent. A hook is an automatic action triggered at a particular point, while a gate is a condition that must pass before work continues.
Skill Claude CodeCodex
A workflow for improving the safeguards and repeatable routines around AI software development. It handles reflection items, checks that block or allow work, and process rules; TDD means writing tests before or alongside code changes.
Skill Claude CodeCodex
A model-tier prompting guide helps write prompts and assign work across AI models with different abilities and speeds. It covers interactive sessions, background agents, and product system prompts.
Skill Claude CodeCodex
Use this skill for plan, architecture, product, technical design, or implementation-plan review when the review should improve the artifact through an adversarial, self-terminating convergence loop. Reviewers attack the plan, an independent verifier tries to refute every blocker before it can block, and a hard…
Skill Claude CodeCodex
A requirement-refinement skill turns a rough development idea into a structured task description. It covers the intent, acceptance criteria, boundaries, missing context, and suggested execution level, then waits for user confirmation.
Skill Claude CodeCodex
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
Agent
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
Agent
Compare two outputs WITHOUT knowing which skill produced them.
Agent
Evaluate expectations against an execution transcript and outputs.
Skill Claude CodeCodex
Use when designing a Claude Code Workflow script for a complex engineering task — multi-agent orchestration, fan-out, adversarial verification, or batch work exceeding a single context. Use before writing any .workflow.js.
Skill Claude CodeCodex
Write or revise a human-readable technical doc (briefing, upgrade notes, announcement, onboarding guide, README, release notes, tutorial) so that a reader with ZERO context can understand AND act on it. Use whenever the user asks to write/refine/review a 宣講文件/升級說明/公告/上手指南/發版說明/briefing/announcement/README, complains a…