展示HN：抗毒性AI代理的QCMP框架（ArXiv Cs.ai待发布）
嘿，HN——经过一年的研究代理人工智能的脆弱性，我构建了QCMP：一个四层架构，旨在防止内存中毒。MCP的服务器数量达到16K，但像MINJA（98.2%的查询成功率）和AgentPoison（从0.1%的毒素中获得80%以上的后门）这样的攻击暴露了核心缺陷——内存过于信任自己。
QCMP借鉴了IIT意识度量（CCI > 0.90以冻结片段）、后量子校验和（ML-KEM-768）、CTC自一致性（NIS > 0.95）以及螳螂虾风格的稀疏检查（<50毫秒TME）。已准备好符合OWASP/EU AI法案，并提供Rust实现的建议。
PDF（浏览器查看）：[https://github.com/bradmcevilly/qcmp-whitepaper/blob/main/QCMP_Whitepaper_arXiv.pdf](https://github.com/bradmcevilly/qcmp-whitepaper/blob/main/QCMP_Whitepaper_arXiv.pdf)
首次向arXiv提交至cs.AI——寻求支持（4个以上的近期订阅）。对量子生物钩子或群体层的反馈如何？欢迎讨论。
deepsweep.ai | linkedin.com/in/bradmcevilly
我在过去一年中专注于解决代理人工智能中的内存中毒问题（例如，仅通过查询就实现98%的MINJA成功率）。介绍QCMP：一个结合IIT意识度量（CCI > 0.90阈值）、后量子校验和（ML-KEM）和CTC一致性的四层架构，以实现防篡改的代理群体。
主要成果：在<50毫秒内检测到0.1%的AgentPoison后门；符合OWASP/EU AI法案。
PDF：[https://github.com/bradmcevilly/qcmp-whitepaper/blob/main/QCMP_Whitepaper_arXiv.pdf](https://github.com/bradmcevilly/qcmp-whitepaper/blob/main/QCMP_Whitepaper_arXiv.pdf)
首次向arXiv提交至cs.AI——寻求HN社区的支持/反馈。对量子生物钩子或多代理层的看法如何？欢迎交流。
网站：deepsweep.ai | LinkedIn：linkedin.com/in/bradmcevilly
查看原文
Hey HN—after a year digging into agentic AI vulnerabilities, I've built QCMP: a 4-layer architecture to slam the door on memory poisoning. MCP's at 16K servers, but attacks like MINJA (98.2% query-only success) and AgentPoison (80%+ backdoors from 0.1% poison) expose the core flaw—memory trusts itself too much.<p>QCMP borrows from IIT consciousness metrics (CCI >0.90 to freeze fragments), post-quantum checksums (ML-KEM-768), CTC self-consistency (NIS >0.95), and mantis shrimp-style sparse checks (<50ms TME). OWASP/EU AI Act ready, with Rust impl tips.<p>PDF (in-browser view): <a href="https://github.com/bradmcevilly/qcmp-whitepaper/blob/main/QCMP_Whitepaper_arXiv.pdf" rel="nofollow">https://github.com/bradmcevilly/qcmp-whitepaper/blob/main/QC...</a><p>First arXiv push to cs.AI—hunting endorsements (4+ recent subs). Feedback on the quantum-bio hooks or swarm layers? Open to riffs.<p>deepsweep.ai | linkedin.com/in/bradmcevilly<p>I've spent the last year tackling memory poisoning in agentic AI (e.g., 98% MINJA success via queries alone). Introducing QCMP: a 4-layer architecture blending IIT consciousness metrics (CCI >0.90 thresholds), post-quantum checksums (ML-KEM), and CTC consistency for tamper-proof agent swarms.<p>Key wins: Detects 0.1% AgentPoison backdoors in <50ms; OWASP/EU AI Act compliant.<p>PDF: <a href="https://github.com/bradmcevilly/qcmp-whitepaper/blob/main/QCMP_Whitepaper_arXiv.pdf" rel="nofollow">https://github.com/bradmcevilly/qcmp-whitepaper/blob/main/QC...</a><p>First arXiv sub to cs.AI—seeking endorsements/feedback from the HN community. Thoughts on the quantum-bio hooks or multi-agent layers? Open to chats.<p>Site: deepsweep.ai | LI: linkedin.com/in/bradmcevilly