by wdnmd1265 · Agent Tool · ★ 35
Two AIs review. A third attacks. You get the truth. Why I built this I started using AI-generated code. The reviews it gave me were confident, but wrong — not obviously wrong. A SQL injection here, a race condition there. Things that looked right at a glance. A single model cannot discover its own blind spots. It reads its output with the same training data, the same biases, the same weak points. It's one person grading their own homework. So I asked: what if two models review independently, and a third actively tries to break the output? Not a vote — an adversarial process. Red team vs.
| Stars | 35 |
| Forks | 3 |
| Language | Python |
| Category | Agent Tool |
| License | Apache-2.0 |
| Quality Score | 69.2477912759254/100 |
| Last Updated | 2026-05-22 |
| Created | 2026-05-18 |
| Platforms | python |
| Est. Tokens | ~293k |
Explore other popular agent tool tools:
ai-flow-architect is Multi-model AI workflow engine with built-in quality arbitration. Plan → Approve → Execute → Audit pipeline. Dual-brain design: Brain #1 (planner) + Brain #2 (arbiter, different model) stops hallucina. It is categorized as a Agent Tool with 35 GitHub stars.
ai-flow-architect is primarily written in Python. It covers topics such as agent, ai, ai-agents.
You can find installation instructions and usage details in the ai-flow-architect GitHub repository at github.com/wdnmd1265/ai-flow-architect. The project has 35 stars and 3 forks, indicating an active community.
ai-flow-architect is released under the Apache-2.0 license, making it free to use and modify according to the license terms.