Claude Code Skills

A collection of custom commands and skills for Claude Code.

Installation

Commands (slash commands)

Symlink the commands/ folder to your Claude Code commands directory:

ln -s /path/to/cc-skills/commands ~/.claude/commands

Skills (auto-triggered)

Symlink the skills/ folder to your Claude skills directory:

ln -s /path/to/cc-skills/skills ~/.claude/skills

This gives you all commands and skills at once, and git pull automatically updates them.

Requirements

codex-review

• OpenAI Codex CLI installed and configured

measuring-pr-performance-impact

• gh CLI authenticated (gh auth login)
• ~/.dogrc with Datadog credentials:

  apikey = your_api_key
  appkey = your_app_key
  

Available Commands

codex-review

Run OpenAI Codex CLI review against a base branch, verify findings, and summarize results.

Usage: /codex-review or /codex-review main

Runs codex review --base <branch>, then manually verifies each finding by examining code and tracing data flow. Produces a structured summary with severity levels (Critical, High, Medium, Low, False Positive).

Available Skills

Skills are auto-triggered based on your request. They include bundled scripts for more reliable execution.

measuring-pr-performance-impact

Measure how a PR affected GraphQL resolver latency using Datadog metrics.

Triggers: "measure PR 27416 performance", "how did PR 27416 affect performance?", "measure PR 27416 impact on portfolio resolver"

Features:

• Latency metrics (avg, p50, p90, p99) and request count
• Manual resolver specification or auto-detection from changed files
• Configurable time window (default 12h)