{"slug": "i-built-a-dual-pool-adversarial-review-system-for-ai-agents-and-it-actually", "title": "I Built a Dual-Pool Adversarial Review System for AI Agents — And It Actually Works", "summary": "A developer built a dual-pool adversarial review system for AI agents that uses real engineers with searchable philosophies instead of abstract roles. The system employs a fixed pool for stability and a random pool for divergence, with managers curating teams per task. Testing on a pull request to the claude-skills repository (18.7K stars) showed the random pool uncovered issues missed by fixed-pool reviewers.", "body_md": "**AI code review has a problem: abstract roles produce generic feedback.** \"Saboteur\" says \"add error handling.\" \"New Hire\" says \"this is confusing.\" Useful? Sometimes. Specific? Rarely.\n\nI built something different: a review system that uses **real engineers with searchable philosophies** instead of abstract roles. Linus Torvalds doesn't say \"consider error handling\" — he says \"eliminate the special case entirely.\" That's not a wording difference. That's a completely different action.\n\n```\nFixed Pool (Convergence)          Random Pool (Divergence)\nDigital-twin matched              Web-searched fresh each time\nStability & depth                 Surprise & blind-spot coverage\n    │                                    │\n    └────────── Cross-orchestrated ──────┘\n              explore ←→ exploit\n```\n\n9 workers + 2 managers, curated to match the user's expertise, personality, and goals. Patty McCord (Netflix's former Chief Talent Officer) and Ed Catmull (Pixar's Braintrust creator) serve as managers who **recruit teams per task** instead of using a fixed template.\n\nFresh personas via web search each session. No preset list — the manager defines search keywords based on what the task needs. This is where the surprises come from.\n\n`[Manager] picked [A,B,C]. Found N issues. Verdict: BLOCK/CONCERNS/CLEAN`\n\nNext round: new manager, keep at most 2 previous members.\n\nI tested this on my own PR to [alirezarezvani/claude-skills](https://github.com/alirezarezvani/claude-skills/pull/866) (18.7K stars):\n\n**The random pool found things both fixed-pool rounds completely missed.** Fixed pool reviewers — who know me — were blind to how an outsider would perceive the skill.\n\n| alirezarezvani adversarial-reviewer | gaurav-yadav adversarial-ai-review | This System | |\n|---|---|---|---|\n| Reviewers | Abstract roles | Domain agents | Real people + searchable philosophy |\n| Team formation | Fixed 3-template | 22 agent pairs | Manager-curated per task |\n| Cross-round | Rotate roles | Same agent set | Swap pool + manager + workers |\n| Personalization | None | None | Digital twin matching |\n| Evolution | Static | Static | Promote/demote/audit cycle |", "url": "https://wpnews.pro/news/i-built-a-dual-pool-adversarial-review-system-for-ai-agents-and-it-actually", "canonical_source": "https://dev.to/yuhaolin2005/i-built-a-dual-pool-adversarial-review-system-for-ai-agents-and-it-actually-works-595j", "published_at": "2026-06-27 14:53:25+00:00", "updated_at": "2026-06-27 15:03:43.926193+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-agents", "developer-tools", "machine-learning", "ai-research"], "entities": ["Linus Torvalds", "Patty McCord", "Ed Catmull", "Netflix", "Pixar", "alirezarezvani", "claude-skills", "gaurav-yadav"], "alternates": {"html": "https://wpnews.pro/news/i-built-a-dual-pool-adversarial-review-system-for-ai-agents-and-it-actually", "markdown": "https://wpnews.pro/news/i-built-a-dual-pool-adversarial-review-system-for-ai-agents-and-it-actually.md", "text": "https://wpnews.pro/news/i-built-a-dual-pool-adversarial-review-system-for-ai-agents-and-it-actually.txt", "jsonld": "https://wpnews.pro/news/i-built-a-dual-pool-adversarial-review-system-for-ai-agents-and-it-actually.jsonld"}}