{
  "markdown": "# Adversarial Review\n\nA Claude Code plugin that orchestrates adversarial review loops between **Claude Code** and **OpenAI Codex CLI**. Claude writes plans and code; Codex tears them apart. They iterate until convergence.\n\n## How It Works\n\nThis plugin adds two slash commands to Claude Code:\n\n### `/adversarial-plan <task description>`\n\nPre-implementation planning with adversarial critique.\n\n1. Claude analyzes your codebase and writes a detailed implementation plan\n2. Codex reviews the plan as a pragmatic senior engineer — looking for real bugs, not theoretical nitpicks\n3. Claude revises based on feedback\n4. Loop continues until Codex approves (or 20 iterations max)\n5. Pauses for your approval before any code is written\n\n### `/adversarial-review`\n\nPost-implementation code review on your current git changes.\n\n1. Captures your git diff\n2. Codex reviews for bugs, security issues, and correctness problems\n3. Claude autonomously fixes any critical issues found\n4. Loop continues until Codex approves (or 20 iterations max)\n\n## Prerequisites\n\n- [Claude Code](https://docs.anthropic.com/en/docs/claude-code) CLI installed and configured\n- [OpenAI Codex CLI](https://github.com/openai/codex) installed and on your PATH (`codex exec` must work)\n\n## Installation\n\nAdd the marketplace and install the plugin in Claude Code:\n\n```\n/plugin marketplace add bobby-beckmann/adversarial-review\n/plugin install adversarial-review@adversarial-review\n```\n\nThe `/adversarial-plan` and `/adversarial-review` commands will be available in all projects.\n\n## Updating\n\n```\n/plugin update adversarial-review@adversarial-review\n```\n\n## Convergence Protocol\n\nCodex ends every review with one of:\n\n- **`APPROVED`** — no blocking issues, work is acceptable\n- **`NEEDS_REVISION`** — blocking issues that must be addressed\n\nReviews are calibrated to be pragmatic. A simple script doesn't get the same scrutiny as a payment system. Convergence typically happens in 2-3 iterations.\n\n## Artifacts\n\nEach run creates a unique session directory under `.adversarial-review/` in your project root:\n\n```\n.adversarial-review/\n  latest -> abcd-efgh-ijkl/       # symlink to most recent session\n  abcd-efgh-ijkl/\n    plan_v1.md, plan_v2.md, ...   # plan iterations\n    review_v1.md, review_v2.md, ... # Codex reviews of each plan\n    plan_final.md                  # converged final plan\n  mnop-qrst-uvwx/\n    diff_for_review.md             # code changes sent for review\n    code_review_v1.md, ...         # Codex code review rounds\n```\n\nPrevious sessions are preserved. You may want to add `.adversarial-review/` to your `.gitignore`.\n\n## Customization\n\n- **Iteration limit**: Edit the commands in `commands/` to change from the default 20\n- **Review prompts**: Customize what Codex evaluates by editing `scripts/codex-review.sh`\n- **Review stance**: The prompts are tuned for pragmatic review — modify them if you want stricter or more lenient feedback\n\n## License\n\nMIT\n",
  "bytes": 2940,
  "sha": "1fc49b7e7cbb0f479a65c6a8f7f20439b6d69049df954728c07ac77e89e105a0",
  "repo_slug": "bobby-beckmann/adversarial-review",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_bobby_beckmann_adversarial_review_advers_bb73c9e8/readme"
}