Local security audit for AI API relays and LLM proxies: detects prompt injection, model substitution, tool-call rewriting, SSE anomalies, error leakage, and Web3 wallet risks.

Adversarial verification framework for AI outputs. Recursive Solver → Auditor → Reconstructor pipeline with benchmark-gated correction logs.

Agent Indoctrination – AI Safety, Bias, Fairness, Ethics & Compliance Testing Framework 🚀

LLM 服务商诚信度检测工具（模型真伪 / token 对账 / 缓存合规 / 性能衰减） · 由 15code 出品

Recon-Level Audit of Claude 4 – Obfuscated, Ethical & Technically Precise

An auditing framework to evaluate LLMs in local government reporting. Compares AI-generated headlines and topic prioritization against professional journalistic standards. Submitted to CHI 2026.

AI agent that transforms existing codebases — no migrations, no rewrites, directly on production code.