The article breaks down the recently leaked source code of Anthropic's Claude Code CLI. It highlights the system's architecture, design choices, and differences from OpenAI's Codex, particularly in handling context overflow and user interactions. Key features like compaction strategies and internal versus external user instructions are explored.

• Claude Code uses a four-tiered compaction strategy (proactive token monitoring, reactive fallback, and a "snip compaction" mode for headless sessions) versus Codex's simpler diff-based approach that just minimizes data sent per turn
• The system prompt uses a boundary marker to cache roughly 3,000 tokens of static instructions across users for performance gains
• Internal users get specialized instructions specifically designed to stop the model from misrepresenting test results or giving misleading status updates
• The codebase includes feature flags and build-time checks specifically to prevent sensitive information from leaking into public builds

This plugin embeds OpenAI Codex into your Claude Code workflow, letting you run standard, adversarial, or rescue reviews without switching tools. Install via Node.js, authenticate with your ChatGPT subscription or API key, then use /codex:review, /codex:adversarial-review, and /codex:rescue alongside status commands.

• Claude Code now has an official Codex plugin (from openai/codex-plugin-cc) with three review modes: standard, adversarial, and rescue
• Adversarial review is meant for high-stakes changes like migrations, auth, or infra scripts to catch subtle flaws standard review might miss
• It runs through your existing Codex CLI/server, so local auth, config, and MCP setup carry over automatically
• An optional "review gate" can force Codex review before Claude Code finishes, but risks tight loops that rapidly burn usage limits