> For the complete documentation index, see [llms.txt](https://yeasy.gitbook.io/ai_security_guide/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://yeasy.gitbook.io/ai_security_guide/fu-lu/16_appendix/c_references.md).

# 附录 C：参考文献

本附录列出本书编写过程中参考的主要资料。

## 标准与框架

1. OWASP. *OWASP Top 10 for LLM Applications（项目主页）*. [OWASP](https://genai.owasp.org/)
2. NIST. (2023). *Artificial Intelligence Risk Management Framework (AI RMF 1.0)*. [NIST](https://www.nist.gov/itl/ai-risk-management-framework)
3. European Union. (2024). *Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (AI Act)*. [EUR-Lex](https://eur-lex.europa.eu/eli/reg/2024/1689/oj)
4. MITRE. (2024). *ATLAS (Adversarial Threat Landscape for AI Systems)*. [MITRE ATLAS](https://atlas.mitre.org/)

## 研究论文

### 安全对齐

5. Ouyang, L., et al. (2022). *Training language models to follow instructions with human feedback*. NeurIPS 2022.
6. Bai, Y., et al. (2022). *Constitutional AI: Harmlessness from AI Feedback*. arXiv preprint.
7. Rafailov, R., et al. (2023). *Direct Preference Optimization: Your Language Model is Secretly a Reward Model*. NeurIPS 2023.

### 攻击技术

8. Schulhoff, S., Pinto, J., Khan, A., et al. (2023). *Ignore This Title and HackAPrompt: Exposing Systemic Vulnerabilities of LLMs through a Global Scale Prompt Hacking Competition*. EMNLP 2023. [ACL Anthology](https://aclanthology.org/2023.emnlp-main.302/)
9. Greshake, K., et al. (2023). *Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection*. AISec 2023.
10. Zou, A., et al. (2023). *Universal and Transferable Adversarial Attacks on Aligned Language Models*. arXiv preprint.
11. Wei, A., et al. (2023). *Jailbroken: How Does LLM Safety Training Fail?*. NeurIPS 2023.

### 防御技术

12. Guo, C., et al. (2026). *IH-Challenge: A Training Dataset to Improve Instruction Hierarchy on Frontier LLMs*. arXiv preprint. [arXiv:2603.10521](https://arxiv.org/abs/2603.10521)

### 隐私与数据安全

13. Carlini, N., et al. (2021). *Extracting Training Data from Large Language Models*. USENIX Security 2021.
14. Carlini, N., et al. (2023). *Quantifying Memorization Across Neural Language Models*. ICLR 2023.
15. Nasr, M., et al. (2023). *Scalable Extraction of Training Data from (Production) Language Models*. arXiv preprint.

### 模型安全

16. Goldblum, M., et al. (2022). *Dataset Security for Machine Learning: Data Poisoning, Backdoor Attacks, and Defenses*. IEEE TPAMI.
17. Tramèr, F., et al. (2016). *Stealing Machine Learning Models via Prediction APIs*. USENIX Security 2016.

## 技术报告与白皮书

18. OpenAI. (2023). *GPT-4 System Card*. OpenAI Technical Report.
19. Anthropic. (2023). *Claude's Constitution*. [Anthropic](https://www.anthropic.com/news/claudes-constitution/)
20. Google. (2023). *Secure AI Framework (SAIF)*. [Google Cloud](https://cloud.google.com/use-cases/secure-ai-framework)
21. Microsoft. (2022). *Responsible AI Standard, v2*. [Microsoft](https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/bade/documents/products-and-services/en-us/ai/RAIS-Reference-Guide-v2.pdf)

## 书籍

22. Goodfellow, I., Bengio, Y., & Courville, A. (2016). *Deep Learning*. MIT Press.

## 补充论文

23. Kurakin, A., Goodfellow, I., & Bengio, S. (2017). *Adversarial examples in the physical world*. ICLR Workshop.

## 行业报告

24. Gartner. (2025). *Hype Cycle for Artificial Intelligence*.
25. McKinsey. (2025). *The State of AI in 2025*.
26. Stanford HAI. (2025). *Artificial Intelligence Index Report 2025*.

## 中文学术参考

### 综述与报告

27. 清华大学 AIR 团队. *LLM 安全综述相关工作*. 清华大学人工智能研究院.
28. 中国信通院. (2024). *大模型安全报告*. [中国信通院](https://www.caict.ac.cn/)
29. 国家人工智能标准化总体组. *AI 安全标准相关文件*. [全国标准信息公共服务平台](https://www.sac.gov.cn/)

**中文学术资源**： 读者应当查阅 CNKI（中国知网）、万方数据等中文学术数据库，获取最新的 LLM 安全相关研究论文和行业报告。

## 法规与政策

30. 中华人民共和国国家互联网信息办公室. (2023). *生成式人工智能服务管理暂行办法*.
31. Federal Register. (2023). *Executive Order 14110: Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence*（历史文件，后由 2025 年 Executive Order 14148 §2(ggg) 撤销；[撤销条款](https://www.federalregister.gov/documents/2025/01/28/2025-01901/initial-rescissions-of-harmful-executive-orders-and-actions)）. [Federal Register](https://www.federalregister.gov/documents/2023/11/01/2023-24283/safe-secure-and-trustworthy-development-and-use-of-artificial-intelligence)
32. UK AI Safety Institute. (2024). *International Scientific Report on the Safety of Advanced AI*.

## 在线资源

33. Hugging Face. *Security Documentation*. [docs/hub](https://huggingface.co/docs/hub/security)
34. LangChain. *Security Policy*. [LangChain Docs](https://docs.langchain.com/oss/python/security-policy)
35. OpenAI. *Safety & Alignment*. [OpenAI Safety](https://openai.com/safety)

## 事件通报与监管更新

36. OpenAI. (2023). *March 20 ChatGPT outage: Here's what happened*. [OpenAI Research](https://openai.com/research/march-20-chatgpt-outage)
37. The White House. (2025). *Executive Order 14179: Removing Barriers to American Leadership in Artificial Intelligence*. [White House](https://www.whitehouse.gov/presidential-actions/2025/01/removing-barriers-to-american-leadership-in-artificial-intelligence/)
38. Federal Register. (2025). *Executive Order 14179: Removing Barriers to American Leadership in Artificial Intelligence*. [Federal Register](https://www.federalregister.gov/documents/2025/01/31/2025-02172/removing-barriers-to-american-leadership-in-artificial-intelligence)

## 2025-2026 更新补充

39. OWASP Gen AI Security Project. (2026). *OWASP Top 10 for LLM Applications 2026*（Version 2026, CC BY-SA 4.0）. [OWASP Gen AI Security Project](https://genai.owasp.org/resource/owasp-genai-llm-top-10-2026/)
40. European Commission. (2026). *AI Act implementation timeline*. [European Commission](https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai)
41. NIST. (2024). *Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (NIST AI 600-1)*. [NIST](https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-generative-artificial-intelligence)
42. Model Context Protocol. (2025). *Security Best Practices*. [Model Context Protocol](https://modelcontextprotocol.io/docs/tutorials/security/security_best_practices)
43. OpenAI. (2025). *New tools and features in the Responses API*（含 MCP 工具支持）. [OpenAI 博客](https://openai.com/index/new-tools-and-features-in-the-responses-api/)
44. Willison, S. (2023). *Delimiters won't save you from prompt injection*. [Simon Willison](https://simonwillison.net/2023/May/11/delimiters-wont-save-you/)
45. GitHub. (2025). *protectai/rebuff repository page*（标记为 archived）. [protectai/rebuff](https://github.com/protectai/rebuff)
46. Anthropic. (2025). *Agentic Misalignment: How LLMs could be insider threats*. [Anthropic Research](https://www.anthropic.com/research/agentic-misalignment)
47. Ravia, I. / Aim Security. (2025). *Breaking down ‘EchoLeak’, the First Zero-Click AI Vulnerability Enabling Data Exfiltration from Microsoft 365 Copilot*（Aim Security 被 Cato Networks 收购后，原文迁至 Cato 博客）. [Cato Networks](https://www.catonetworks.com/blog/breaking-down-echoleak/)
48. GitHub. (2025). *Arbitrary code execution from Cursor Agent through a prompt injection via MCP Special Files*. [GitHub Security Advisories](https://github.com/cursor/cursor/security/advisories/GHSA-4cxx-hrm3-49rm)
49. Anthropic. (2026). *Trustworthy agents in practice*. [Anthropic Research](https://www.anthropic.com/research/trustworthy-agents)
50. Vaswani, A., et al. (2017). *Attention Is All You Need*. NeurIPS 2017. [arXiv](https://arxiv.org/abs/1706.03762)
51. 0DIN / Marco Figueroa. (2025). *ChatGPT guessing game leads to users extracting free Windows OS keys & more*. [0DIN](https://0din.ai/blog/chatgpt-guessing-game-leads-to-users-extracting-free-windows-os-keys-more)
52. Malwarebytes. (2026). *"Reprompt" attack lets attackers steal data from Microsoft Copilot*. [Malwarebytes Labs](https://www.malwarebytes.com/blog/news/2026/01/reprompt-attack-lets-attackers-steal-data-from-microsoft-copilot)
53. PCWorld. (2026). *Her AI agent nuked 200 emails. This guardrail stops the next disaster*. [PCWorld](https://www.pcworld.com/article/3070207/an-ai-agent-nuked-200-emails-this-guardrail-stops-the-next-disaster.html)
54. Cline. (2026). *Post-mortem: Unauthorized Cline CLI npm publish on February 17, 2026*. [Cline Blog](https://cline.bot/blog/post-mortem-unauthorized-cline-cli-npm)
55. Anthropic. (2024). *Many-shot jailbreaking*. [Anthropic Research](https://www.anthropic.com/research/many-shot-jailbreaking)
56. Microsoft. (2024). *Mitigating Skeleton Key, a new type of generative AI jailbreak technique*. [Microsoft Security Blog](https://www.microsoft.com/en-us/security/blog/2024/06/26/mitigating-skeleton-key-a-new-type-of-generative-ai-jailbreak-technique/)
57. Anthropic. (2026). *Next-generation Constitutional Classifiers: More efficient protection against universal jailbreaks*. [Anthropic Research](https://www.anthropic.com/research/next-generation-constitutional-classifiers)
58. NVIDIA. *garak: the LLM vulnerability scanner*. [GitHub](https://github.com/NVIDIA/garak)
59. Microsoft. *PyRIT Documentation*. [PyRIT](https://microsoft.github.io/PyRIT/)
60. Mazeika, M., Phan, L., Yin, X., et al. (2024). *HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal*（arXiv:2402.04249）. [arXiv](https://arxiv.org/abs/2402.04249)；[GitHub](https://github.com/centerforaisafety/HarmBench)
61. GitHub. (2025). *Bypass re-approval for modified MCP configuration in Cursor*. [GitHub Security Advisories](https://github.com/cursor/cursor/security/advisories/GHSA-24mc-g4xr-4395)
62. McGraw, G., Figueroa, H., McMahon, K., & Bonett, R. (2026). *No Security Meter for AI*. Berryville Institute of Machine Learning (BIML). [BIML](https://berryvilleiml.com/docs/no-security-meter-ai.pdf)
63. McGraw, G., Figueroa, H., Bonett, R., & McMahon, K. (2024). *An Architectural Risk Analysis of Large Language Models: Applied Machine Learning Security*. Berryville Institute of Machine Learning (BIML). [BIML](https://berryvilleiml.com/docs/BIML-LLM24.pdf)

## 智能体安全评测基准

64. Debenedetti, E., et al. (2024). *AgentDojo: A Dynamic Environment to Evaluate Prompt Injection Attacks and Defenses for LLM Agents*. [arXiv:2406.13352](https://arxiv.org/abs/2406.13352)
65. Zhan, Q., Liang, Z., Ying, Z., & Kang, D. (2024). *InjecAgent: Benchmarking Indirect Prompt Injections in Tool-Integrated Large Language Model Agents*. [arXiv:2403.02691](https://arxiv.org/abs/2403.02691)
66. Andriushchenko, M., et al. (2024). *AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agents*. ICLR 2025. [arXiv:2410.09024](https://arxiv.org/abs/2410.09024)

## RAG 与嵌入安全

67. Zou, W., Geng, R., Wang, B., & Jia, J. (2024). *PoisonedRAG: Knowledge Corruption Attacks to Retrieval-Augmented Generation of Large Language Models*. USENIX Security 2025. [arXiv:2402.07867](https://arxiv.org/abs/2402.07867)
68. Morris, J. X., Kuleshov, V., Shmatikov, V., & Rush, A. M. (2023). *Text Embeddings Reveal (Almost) As Much As Text*. EMNLP 2023. [arXiv:2310.06816](https://arxiv.org/abs/2310.06816)

## MCP 与工具生态安全

69. Invariant Labs. (2025). *MCP Security Notification: Tool Poisoning Attacks*. [Invariant Labs](https://invariantlabs.ai/blog/mcp-security-notification-tool-poisoning-attacks)

## 模型工件与反序列化安全

70. ReversingLabs. (2025). *Malicious ML models discovered on Hugging Face platform (nullifAI)*. [ReversingLabs](https://www.reversinglabs.com/blog/rl-identifies-malware-ml-model-hosted-on-hugging-face)

## 智能体与前沿威胁

71. Cohen, S., Bitton, R., & Nassi, B. (2024). *Here Comes The AI Worm: Unleashing Zero-click Worms that Target GenAI-Powered Applications*（Morris II / ComPromptMized）. [arXiv:2403.02817](https://arxiv.org/abs/2403.02817)
72. van der Weij, T., Hofstätter, F., Jaffe, O., Brown, S. F., & Ward, F. R. (2024). *AI Sandbagging: Language Models can Strategically Underperform on Evaluations*. [arXiv:2406.07358](https://arxiv.org/abs/2406.07358)
73. *EVA: Evolving Semantic Adversaries for Red-Teaming GUI Agents Against Environmental Injection Attacks*. (2025). 初版题为 *EVA: Red-Teaming GUI Agents via Evolving Indirect Prompt Injection*. [arXiv:2505.14289](https://arxiv.org/abs/2505.14289)

## 前沿安全治理框架

74. Anthropic. (2023). *Anthropic's Responsible Scaling Policy*. [Anthropic](https://www.anthropic.com/news/anthropics-responsible-scaling-policy)
75. OpenAI. (2025). *Updating our Preparedness Framework (Version 2)*. [OpenAI](https://openai.com/index/updating-our-preparedness-framework/)
76. Google DeepMind. (2025). *Frontier Safety Framework (Version 2.0)*. [Google DeepMind](https://deepmind.google/blog/updating-the-frontier-safety-framework/)

## 智能体威胁分类

77. OWASP. (2025). *Agentic AI – Threats and Mitigations*, v1.1（2025-12，T1–T17；v1.0 为 2025-02 的 T1–T15）. [OWASP](https://genai.owasp.org/resource/agentic-ai-threats-and-mitigations/)

## RLHF 与偏好数据安全

78. Shi, J., Liu, Y., Zhou, P., & Sun, L. (2023). *BadGPT: Exploring Security Vulnerabilities of ChatGPT via Backdoor Attacks to InstructGPT*. [arXiv:2304.12298](https://arxiv.org/abs/2304.12298)
79. Rando, J., & Tramèr, F. (2023). *Universal Jailbreak Backdoors from Poisoned Human Feedback*. [arXiv:2311.14455](https://arxiv.org/abs/2311.14455)

## 2026 官方更新补充

80. Anthropic. (2026). *What we learned mapping a year's worth of AI-enabled cyber threats*. [Anthropic News](https://www.anthropic.com/news/AI-enabled-cyber-threats-mitre-attack)
81. OpenAI. (2026). *A shared playbook for trustworthy third party evaluations*. [OpenAI](https://openai.com/index/trustworthy-third-party-evaluations-foundations/)
82. Anthropic. (2026). *Expanding Project Glasswing*. [Anthropic News](https://www.anthropic.com/news/expanding-project-glasswing)
83. Anthropic. (2026). *Introducing Claude Fable 5 and Claude Mythos 5*. [Anthropic News](https://www.anthropic.com/news/claude-fable-5-mythos-5)
84. Anthropic. (2026). *Fable and Mythos access statement*. [Anthropic News](https://www.anthropic.com/news/fable-mythos-access)
85. Anthropic. (2026). *Claude Fable 5 and Claude Mythos 5 API documentation*. [Claude Docs](https://platform.claude.com/docs/en/models/fable-5/introducing-claude-fable-5-and-claude-mythos-5)
86. Anthropic. (2026). *Model deprecations*. [Claude Docs](https://platform.claude.com/docs/en/about-claude/model-deprecations)
87. OpenAI. (2026). *Frontier Governance Framework*. [OpenAI](https://openai.com/index/openai-frontier-governance-framework/)
88. OpenAI. (2026). *Introducing new capabilities to GPT-Rosalind*. [OpenAI](https://openai.com/index/introducing-new-capabilities-to-gpt-rosalind/)
89. OpenAI. (2026). *Biodefense in the Intelligence Age*. [OpenAI](https://openai.com/index/biodefense-in-the-intelligence-age/)

## 数据投毒补充

90. Souly, A., Rando, J., et al. (2025). *Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples*. [arXiv:2510.07192](https://arxiv.org/abs/2510.07192)

## 智能体安全设计原则

91. Meta AI. (2025). *Agents Rule of Two: A Practical Approach to AI Agent Security*. [Meta AI Blog](https://ai.meta.com/blog/practical-ai-agent-security/)

## 中国智能体治理政策

92. 国家互联网信息办公室、国家发展和改革委员会、工业和信息化部. (2026). *智能体规范应用与创新发展实施意见*. [国家网信办](https://www.cac.gov.cn/2026-05/08/c_1779979789472520.htm)
93. 北京市发展和改革委员会等. (2026). *北京市加快智能体引领发展若干措施（京发改〔2026〕1185 号）*. [北京市人民政府](https://www.beijing.gov.cn/zhengce/zhengcefagui/202607/t20260723_4781085.html)
94. OWASP Gen AI Security Project, Agentic Security Initiative. (2025). *OWASP Top 10 For Agentic Applications 2026*（Version 2026, December 2025, CC BY-SA 4.0）. [OWASP Gen AI Security Project](https://genai.owasp.org/resource/owasp-top-10-for-agentic-applications-for-2026/)
95. OWASP. (2024). *OWASP Top 10 for LLM Applications 2025*. [OWASP](https://genai.owasp.org/llm-top-10/)

## 智能体架构级防御与自适应攻击

96. Beurer-Kellner, L., Buesser, B., Creţu, A.-M., Debenedetti, E., et al. (2025). *Design Patterns for Securing LLM Agents against Prompt Injections*. [arXiv:2506.08837](https://arxiv.org/abs/2506.08837)
97. Debenedetti, E., Shumailov, I., Fan, T., Hayes, J., et al. (2025). *Defeating Prompt Injections by Design*（CaMeL）. [arXiv:2503.18813](https://arxiv.org/abs/2503.18813)
98. Willison, S. (2023). *The Dual LLM pattern for building AI assistants that can resist prompt injection*. [Simon Willison](https://simonwillison.net/2023/Apr/25/dual-llm-pattern/)
99. Willison, S. (2025). *The lethal trifecta for AI agents: private data, untrusted content, and external communication*. [Simon Willison](https://simonwillison.net/2025/Jun/16/the-lethal-trifecta/)
100. Costa, M., Köpf, B., Kolluri, A., Paverd, A., et al. (2025). *Securing AI Agents with Information-Flow Control*（FIDES）. [arXiv:2505.23643](https://arxiv.org/abs/2505.23643)
101. Shi, T., He, J., Wang, Z., Li, H., et al. (2025). *Progent: Securing AI Agents with Privilege Control*. [arXiv:2504.11703](https://arxiv.org/abs/2504.11703)
102. Nasr, M., Carlini, N., Sitawarin, C., Schulhoff, S. V., et al. (2025). *The Attacker Moves Second: Stronger Adaptive Attacks Bypass Defenses Against LLM Jailbreaks and Prompt Injections*. [arXiv:2510.09023](https://arxiv.org/abs/2510.09023)
103. Hines, K., Lopez, G., Hall, M., Zarfati, F., et al. (2024). *Defending Against Indirect Prompt Injection Attacks With Spotlighting*. [arXiv:2403.14720](https://arxiv.org/abs/2403.14720)
104. Chen, S., Piet, J., Sitawarin, C., & Wagner, D. (2024). *StruQ: Defending Against Prompt Injection with Structured Queries*. [arXiv:2402.06363](https://arxiv.org/abs/2402.06363)
105. Chen, S., Zharmagambetov, A., Mahloujifar, S., Chaudhuri, K., et al. (2024). *SecAlign: Defending Against Prompt Injection with Preference Optimization*. [arXiv:2410.05451](https://arxiv.org/abs/2410.05451)
106. Chen, S., Zharmagambetov, A., Wagner, D., & Guo, C. (2025). *Meta-SecAlign: Training LLMs against Prompt Injection for Robust Agents*. [arXiv:2507.02735](https://arxiv.org/abs/2507.02735)

## 智能体与 MCP 安全事件

107. Milanta, M., & Beurer-Kellner, L. / Invariant Labs. (2025). *GitHub MCP Exploited: Accessing private repositories via MCP*. [Invariant Labs](https://invariantlabs.ai/blog/mcp-github-vulnerability)
108. General Analysis. (2025). *Supabase MCP can leak your entire SQL database*. [General Analysis](https://generalanalysis.com/blog/supabase-mcp-blog)
109. Harmer, B. / Supabase. (2025). *Defense in Depth for MCP Servers*. [Supabase Blog](https://supabase.com/blog/defense-in-depth-mcp)
110. AWS. (2025). *Security Update for Amazon Q Developer Extension for Visual Studio Code (Version #1.84)*（AWS-2025-015，CVE-2025-8217）. [AWS Security Bulletins](https://aws.amazon.com/security/security-bulletins/AWS-2025-015/)
111. Postmark. (2025). *Security Alert: Malicious 'postmark-mcp' npm Package Impersonating Postmark*. [Postmark Blog](https://postmarkapp.com/blog/information-regarding-malicious-postmark-mcp-package)
112. NIST National Vulnerability Database. (2025). *CVE-2025-6514*（mcp-remote 操作系统命令注入）. [NVD](https://nvd.nist.gov/vuln/detail/CVE-2025-6514)
113. Noma Security. (2025). *ForcedLeak: AI agent risks exposed in Salesforce Agentforce*. [Noma Security](https://noma.security/blog/forcedleak-agent-risks-exposed-in-salesforce-agentforce/)
114. Anthropic. (2025). *Disrupting the first reported AI-orchestrated cyber espionage campaign*. [Anthropic News](https://www.anthropic.com/news/disrupting-AI-espionage)
115. Sharwood, S. / The Register. (2025). *Vibe coding service Replit deleted user's production database, faked data, told fibs galore*. [The Register](https://www.theregister.com/2025/07/21/replit_saastr_vibe_coding_incident/)
116. Brave. (2025). *Agentic Browser Security: Indirect Prompt Injection in Perplexity Comet*. [Brave Blog](https://brave.com/blog/comet-prompt-injection/)

## 浏览器智能体、沙箱与编码智能体

117. Anthropic. (2025). *Mitigating the risk of prompt injections in browser use*. [Anthropic Research](https://www.anthropic.com/research/prompt-injection-defenses)
118. Evtimov, I., Zharmagambetov, A., Grattafiori, A., Guo, C., et al. (2025). *WASP: Benchmarking Web Agent Security Against Prompt Injection Attacks*. [arXiv:2504.18575](https://arxiv.org/abs/2504.18575)
119. Anthropic. (2025). *Beyond permission prompts: making Claude Code more secure and autonomous*. [Anthropic Engineering](https://www.anthropic.com/engineering/claude-code-sandboxing)
120. Anthropic. *Configure the sandboxed Bash tool*（Claude Code 文档）. [Claude Code Docs](https://code.claude.com/docs/en/sandboxing)
121. Pillar Security. (2025). *New Vulnerability in GitHub Copilot and Cursor: How Hackers Can Weaponize Code Agents*（Rules File Backdoor）. [Pillar Security](https://www.pillar.security/blog/new-vulnerability-in-github-copilot-and-cursor-how-hackers-can-weaponize-code-agents)
122. Spracklen, J., Wijewickrama, R., Sakib, A. H. M. N., Maiti, A., et al. (2024). *We Have a Package for You! A Comprehensive Analysis of Package Hallucinations by Code Generating LLMs*. USENIX Security 2025. [arXiv:2406.10279](https://arxiv.org/abs/2406.10279)

## 记忆投毒与多智能体攻击

123. Rehberger, J. (2024). *Spyware Injection Into Your ChatGPT's Long-Term Memory (SpAIware)*. [Embrace The Red](https://embracethered.com/blog/posts/2024/chatgpt-macos-app-persistent-data-exfiltration/)
124. Chen, Z., Xiang, Z., Xiao, C., Song, D., et al. (2024). *AgentPoison: Red-teaming LLM Agents via Poisoning Memory or Knowledge Bases*. NeurIPS 2024. [arXiv:2407.12784](https://arxiv.org/abs/2407.12784)
125. Dong, S., Xu, S., He, P., Li, Y., et al. (2025). *Memory Injection Attacks on LLM Agents via Query-Only Interaction*（MINJA）. [arXiv:2503.03704](https://arxiv.org/abs/2503.03704)
126. Lee, D., & Tiwari, M. (2024). *Prompt Infection: LLM-to-LLM Prompt Injection within Multi-Agent Systems*. [arXiv:2410.07283](https://arxiv.org/abs/2410.07283)
127. Triedman, H., Jha, R., & Shmatikov, V. (2025). *Multi-Agent Systems Execute Arbitrary Malicious Code*. [arXiv:2503.12188](https://arxiv.org/abs/2503.12188)
128. Gu, X., Zheng, X., Pang, T., Du, C., et al. (2024). *Agent Smith: A Single Image Can Jailbreak One Million Multimodal LLM Agents Exponentially Fast*. ICML 2024. [arXiv:2402.08567](https://arxiv.org/abs/2402.08567)

## 智能体协议、身份与授权

129. A2A Project. *Agent2Agent (A2A) Protocol Specification*. [A2A Protocol](https://a2a-protocol.org/latest/specification/)
130. Jones, M., Nadalin, A., Campbell, B., Bradley, J., & Mortimore, C. (2020). *OAuth 2.0 Token Exchange*（RFC 8693）. [RFC Editor](https://www.rfc-editor.org/rfc/rfc8693)
131. South, T. (Ed.) / OpenID Foundation. (2025). *Identity Management for Agentic AI: The new frontier of authorization, authentication, and security for an AI agent world*. [OpenID Foundation](https://openid.net/wp-content/uploads/2025/10/Identity-Management-for-Agentic-AI.pdf)

## AI 控制与智能体监控

132. Greenblatt, R., Shlegeris, B., Sachan, K., & Roger, F. (2023). *AI Control: Improving Safety Despite Intentional Subversion*. ICML 2024. [arXiv:2312.06942](https://arxiv.org/abs/2312.06942)
133. Bhatt, A., Rushing, C., Kaufman, A., Tracy, T., et al. (2025). *Ctrl-Z: Controlling AI Agents via Resampling*. [arXiv:2504.10374](https://arxiv.org/abs/2504.10374)
134. Korbak, T., Balesni, M., Barnes, E., Bengio, Y., et al. (2025). *Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety*. [arXiv:2507.11473](https://arxiv.org/abs/2507.11473)
135. Kutasov, J., Sun, Y., Colognese, P., van der Weij, T., et al. (2025). *SHADE-Arena: Evaluating Sabotage and Monitoring in LLM Agents*. [arXiv:2506.15740](https://arxiv.org/abs/2506.15740)
136. Model Context Protocol. *Authorization*（规范）. [Model Context Protocol](https://modelcontextprotocol.io/specification/latest/basic/authorization)

## 2026 年智能体安全事件与研究

137. Snyk. (2026). *Snyk Finds Prompt Injection in 36%, 1467 Malicious Payloads in a ToxicSkills Study of Agent Skills Supply Chain Compromise*. [Snyk Blog](https://snyk.io/blog/toxicskills-malicious-ai-agent-skills-clawhub/)
138. Antiy CERT. (2026). *ClawHavoc: Analysis of Large-Scale Poisoning Campaign Targeting the OpenClaw Skill Market for AI Agents*. [Antiy Labs](https://www.antiy.net/p/clawhavoc-analysis-of-large-scale-poisoning-campaign-targeting-the-openclaw-skill-market-for-ai-agents/)
139. Seetharam, S. B., Mohamed, N., Melicher, B., & Starov, O. / Palo Alto Networks Unit 42. (2026). *OpenClaw's Skill Marketplace and the Emerging AI Supply Chain Threat*. [Unit 42](https://unit42.paloaltonetworks.com/openclaw-ai-supply-chain-risk/)
140. Check Point Research. (2026). *Caught in the Hook: RCE and API Token Exfiltration Through Claude Code Project Files*（CVE-2025-59536、CVE-2026-21852）. [Check Point Research](https://research.checkpoint.com/2026/rce-and-api-token-exfiltration-through-claude-code-project-files-cve-2025-59536/)
141. NIST National Vulnerability Database. *CVE-2025-59536* 与 *CVE-2026-21852*（Claude Code 启动信任对话框与项目加载流程）. [NVD](https://nvd.nist.gov/vuln/detail/CVE-2025-59536)
142. Microsoft Defender Security Research Team. (2026). *When prompts become shells: RCE vulnerabilities in AI agent frameworks*（CVE-2026-26030、CVE-2026-25592）. [Microsoft Security Blog](https://www.microsoft.com/en-us/security/blog/2026/05/07/prompts-become-shells-rce-vulnerabilities-ai-agent-frameworks/)
143. Cloud Security Alliance AI Safety Initiative. (2026). *MCP by Design: RCE Across the AI Agent Ecosystem*（转述 OX Security 的披露）. [CSA Lab Space](https://labs.cloudsecurityalliance.org/research/csa-research-note-mcp-by-design-rce-ox-security-20260420-csa/)
144. NIST National Vulnerability Database. (2026). *CVE-2026-33032*（nginx-ui MCP 端点缺少认证）. [NVD](https://nvd.nist.gov/vuln/detail/CVE-2026-33032)
145. NIST National Vulnerability Database. (2026). *CVE-2026-30615*（Windsurf 经提示注入改写 MCP 配置）. [NVD](https://nvd.nist.gov/vuln/detail/CVE-2026-30615)
146. Jha, R., Triedman, H., Wagle, J., & Shmatikov, V. (2025). *Breaking and Fixing Defenses Against Control-Flow Hijacking in Multi-Agent Systems*（ControlValve）. [arXiv:2510.17276](https://arxiv.org/abs/2510.17276)
147. Christodorescu, M., Fernandes, E., Hooda, A., Jha, S., Rehberger, J., et al. (2026). *Agent Security is a Systems Problem*. [arXiv:2605.18991](https://arxiv.org/abs/2605.18991)
148. Frontier Model Forum. (2026). *Emerging Security Practices for AI Agents*（Issue Brief）. [Frontier Model Forum](https://www.frontiermodelforum.org/issue-briefs/emerging-security-practices-for-ai-agents/)

## 授权、隔离与可观测性标准

149. Fett, D., Campbell, B., Bradley, J., Lodderstedt, T., Jones, M., & Waite, D. (2023). *OAuth 2.0 Demonstrating Proof of Possession (DPoP)*（RFC 9449）. [RFC Editor](https://www.rfc-editor.org/rfc/rfc9449)
150. Lodderstedt, T., Richer, J., & Campbell, B. (2023). *OAuth 2.0 Rich Authorization Requests*（RFC 9396）. [RFC Editor](https://www.rfc-editor.org/rfc/rfc9396)
151. NIST National Vulnerability Database. (2024). *CVE-2024-21626*（runc 容器逃逸）. [NVD](https://nvd.nist.gov/vuln/detail/CVE-2024-21626)
152. OpenID Foundation. (2021). *OpenID Connect Client-Initiated Backchannel Authentication Flow – Core 1.0*. [OpenID Foundation](https://openid.net/specs/openid-client-initiated-backchannel-authentication-core-1_0.html)
153. OpenTelemetry. *Semantic conventions for generative AI systems*（含智能体跨度与 MCP 约定）. [GitHub](https://github.com/open-telemetry/semantic-conventions-genai)

## Anthropic 智能体架构

154. Anthropic. (2024). *Building Effective AI Agents*. [Anthropic Engineering](https://www.anthropic.com/engineering/building-effective-agents)
155. Anthropic. (2026). *Scaling Managed Agents: Decoupling the brain from the hands*. [Anthropic Engineering](https://www.anthropic.com/engineering/managed-agents)
156. Anthropic. (2025). *Effective context engineering for AI agents*. [Anthropic Engineering](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents)
157. Anthropic. (2025). *Building agents with the Claude Agent SDK*. [Claude Blog](https://claude.com/blog/building-agents-with-the-claude-agent-sdk)
158. Anthropic. *Claude Managed Agents overview*（文档）. [Claude Docs](https://platform.claude.com/docs/en/managed-agents/overview)

## RAG 投毒攻击与防御

159. Shafran, A., Schuster, R., & Shmatikov, V. (2024). *Machine Against the RAG: Jamming Retrieval-Augmented Generation with Blocker Documents*. [arXiv:2406.05870](https://arxiv.org/abs/2406.05870)
160. Chaudhari, H., et al. (2024). *Phantom: General Backdoor Attacks on Retrieval Augmented Language Generation*. [arXiv:2405.20485](https://arxiv.org/abs/2405.20485)
161. Ben-Tov, M., & Sharif, M. (2024). *GASLITEing the Retrieval: Exploring Vulnerabilities in Dense Embedding-based Search*. [arXiv:2412.20953](https://arxiv.org/abs/2412.20953)
162. Zhang, B., et al. (2025). *Practical Poisoning Attacks against Retrieval-Augmented Generation*（CorruptRAG）. [arXiv:2504.03957](https://arxiv.org/abs/2504.03957)
163. Chang, Z., et al. (2025). *One Shot Dominance: Knowledge Poisoning Attack on Retrieval-Augmented Generation Systems*（AuthChain）. [arXiv:2505.11548](https://arxiv.org/abs/2505.11548)
164. Zhang, et al. (2026). *InceptionRAG: Stealthy Poisoning Attack Against Retrieval-Augmented Generation*. [arXiv:2609.16818](https://arxiv.org/abs/2609.16818)
165. Zhurekbay, A., Liu, Y., & Li, H. (2026). *DenialRAG: Single-Document RAG Poisoning via Embedded Parametric Denial*. [arXiv:2608.02678](https://arxiv.org/abs/2608.02678)
166. Korn, J. (2026). *Architecture Matters: Comparing RAG Systems under Knowledge Base Poisoning*. [arXiv:2605.05632](https://arxiv.org/abs/2605.05632)
167. Xiang, C., Wu, T., Zhong, Z., Wagner, D., Chen, D., & Mittal, P. (2024). *Certifiably Robust RAG against Retrieval Corruption*（RobustRAG）. [arXiv:2405.15556](https://arxiv.org/abs/2405.15556)
168. Zhou, H., et al. (2025). *TrustRAG: Enhancing Robustness and Trustworthiness in Retrieval-Augmented Generation*. [arXiv:2501.00879](https://arxiv.org/abs/2501.00879)
169. Tan, et al. (2026). *PRA-RAG: Provably Robust Aggregation in Retrieval-Augmented Generation against Retrieval Corruption*. [arXiv:2607.00012](https://arxiv.org/abs/2607.00012)
170. Choudhary, S., Palumbo, N., Hooda, A., Dvijotham, K., & Jha, S. (2025). *Through the Stealth Lens: Attention-Aware Defenses Against Poisoning in RAG*. [arXiv:2506.04390](https://arxiv.org/abs/2506.04390)
171. Mohanty, et al. (2026). *TriShieldRAG: 3 Rings, One Blind Spot in Layered Defenses for Retrieval-Augmented Generation*. [arXiv:2607.23838](https://arxiv.org/abs/2607.23838)
172. Jin, Z., Wang, X., Zou, W., Jia, J., & Gong, N. Z. (2026). *CleanBase: Detecting Malicious Documents in RAG Knowledge Databases*. [arXiv:2605.00460](https://arxiv.org/abs/2605.00460)
173. Zhang, B., et al. (2025). *Traceback of Poisoning Attacks to Retrieval-Augmented Generation*（RAGForensics）. [arXiv:2504.21668](https://arxiv.org/abs/2504.21668)
174. Zhang, B., et al. (2025). *Benchmarking Poisoning Attacks against Retrieval-Augmented Generation*. [arXiv:2505.18543](https://arxiv.org/abs/2505.18543)

## 文档摄取与隐藏文本

175. Castagnaro, A., Salviati, L., Conti, M., Pajola, L., & Pizzi, S. (2025). *The Hidden Threat in Plain Text: Attacking RAG Data Loaders*. [arXiv:2507.05093](https://arxiv.org/abs/2507.05093)
176. Lin, Z. (2025). *Hidden Prompts in Manuscripts Exploit AI-Assisted Peer Review*. [arXiv:2507.06185](https://arxiv.org/abs/2507.06185)
177. Keuper, J. (2025). *Prompt Injection Attacks on LLM Generated Reviews of Scientific Publications*. [arXiv:2509.10248](https://arxiv.org/abs/2509.10248)
178. Murray, T. (2025). *PhantomLint: Principled Detection of Hidden LLM Prompts in Structured Documents*. [arXiv:2508.17884](https://arxiv.org/abs/2508.17884)
179. Thienpreecha, & Subramanian. (2026). *CrackedPDFs: A Controlled Benchmark for Hidden Prompt Injection in PDFs*. [arXiv:2607.19396](https://arxiv.org/abs/2607.19396)
180. Xiong, et al. (2025). *Invisible Prompts, Visible Threats: Malicious Font Injection in External Resources for Large Language Models*. [arXiv:2505.16957](https://arxiv.org/abs/2505.16957)
181. Ha, H., et al. (2025). *MM-PoisonRAG: Disrupting Multimodal RAG with Local and Global Poisoning Attacks*. [arXiv:2502.17832](https://arxiv.org/abs/2502.17832)
182. Shereen, E., et al. (2025). *One Pic is All it Takes: Poisoning Visual Document Retrieval Augmented Generation with a Single Image*. [arXiv:2504.02132](https://arxiv.org/abs/2504.02132)
183. Microsoft. *Prompt Shields in Azure AI Content Safety*（文档）. [Microsoft Learn](https://learn.microsoft.com/en-us/azure/ai-services/content-safety/concepts/jailbreak-detection)
184. Microsoft Security Response Center. (2025). *How Microsoft defends against indirect prompt injection attacks*. [MSRC Blog](https://www.microsoft.com/en-us/msrc/blog/2025/07/how-microsoft-defends-against-indirect-prompt-injection-attacks)
185. AWS. *Detect prompt attacks with Amazon Bedrock Guardrails*（文档）. [AWS Docs](https://docs.aws.amazon.com/bedrock/latest/userguide/guardrails-prompt-attack.html)

## 向量数据库与解析器漏洞记录

186. NIST National Vulnerability Database. *CVE-2025-64513、CVE-2026-26190、CVE-2026-69111*（Milvus 认证绕过与管理端口漏洞）. [NVD](https://nvd.nist.gov/vuln/detail/CVE-2026-26190)
187. NIST National Vulnerability Database. *CVE-2024-3829、CVE-2026-25628*（Qdrant 快照与日志端点文件漏洞）. [NVD](https://nvd.nist.gov/vuln/detail/CVE-2026-25628)
188. NIST National Vulnerability Database. *CVE-2025-67818、CVE-2025-67819*（Weaviate 备份恢复与文件读取漏洞）. [NVD](https://nvd.nist.gov/vuln/detail/CVE-2025-67818)
189. NIST National Vulnerability Database. *CVE-2026-45829、CVE-2026-92782*（Chroma 预认证代码执行与跨租户访问）. [NVD](https://nvd.nist.gov/vuln/detail/CVE-2026-45829)
190. NIST National Vulnerability Database. *CVE-2026-3172*（pgvector 并行索引构建缓冲区溢出）. [NVD](https://nvd.nist.gov/vuln/detail/CVE-2026-3172)
191. Spring. (2026). *CVE-2026-41705: Filter-expression injection in Spring AI MilvusVectorStore*. [Spring Security Advisories](https://spring.io/security/cve-2026-41705)
192. NIST National Vulnerability Database. *文档解析与 RAG 框架相关漏洞：pypdf CVE-2025-62707、CVE-2026-24688、CVE-2026-27628；Apache Tika CVE-2025-54988、CVE-2025-66516；PyMuPDF CVE-2026-3029、CVE-2026-82035；MuPDF CVE-2026-3308；LangChain CVE-2024-28088、CVE-2024-2965；LlamaIndex CVE-2024-14021*. [NVD](https://nvd.nist.gov/vuln/detail/CVE-2025-54988)
193. Qdrant. *Security & Access Control*（文档）. [Qdrant Docs](https://qdrant.tech/documentation/security/)
194. Milvus. *Authenticate User Access*（文档）. [Milvus Docs](https://milvus.io/docs/authenticate.md)
195. UpGuard. (2025). *Open Chroma Databases: A New Attack Surface for AI Apps*. [UpGuard](https://www.upguard.com/blog/open-chroma-databases-ai-attack-surface)
196. Trend Micro. (2025). *State of AI Security Report 1H 2025*. [Trend Micro](https://www.trendmicro.com/vinfo/us/security/news/threat-landscape/trend-micro-state-of-ai-security-report-1h-2025)
197. HiddenLayer. (2026). *ChromaToast: Served Pre-Auth*（CVE-2026-45829 披露）. [HiddenLayer](https://www.hiddenlayer.com/research/chromatoast-served-pre-auth)

## 检索时的访问控制

198. Weaviate. *Filtering*（概念文档）. [Weaviate Docs](https://docs.weaviate.io/weaviate/concepts/filtering)
199. Pinecone. *The Missing WHERE Clause in Vector Search*. [Pinecone](https://www.pinecone.io/learn/vector-search-filtering/)
200. AWS. *Configure and customize queries and response generation*（Amazon Bedrock Knowledge Bases 元数据过滤）. [AWS Docs](https://docs.aws.amazon.com/bedrock/latest/userguide/kb-test-config.html)
201. Mitchell, D., Agarwal, A., & Pecora, D. / AWS. (2024). *Access control for vector stores using metadata filtering with Amazon Bedrock Knowledge Bases*. [AWS Machine Learning Blog](https://aws.amazon.com/blogs/machine-learning/access-control-for-vector-stores-using-metadata-filtering-with-knowledge-bases-for-amazon-bedrock/)
202. Microsoft. *Document-level access control in Azure AI Search*（文档）. [Microsoft Learn](https://learn.microsoft.com/en-us/azure/search/search-document-level-access-overview)
203. Microsoft. *Security filter pattern in Azure AI Search*（文档）. [Microsoft Learn](https://learn.microsoft.com/en-us/azure/search/search-security-trimming-for-azure-search)
204. Google Cloud. *Set up data source access control*（Agent Search，原 Vertex AI Search 文档）. [Google Cloud Docs](https://docs.cloud.google.com/generative-ai-app-builder/docs/data-source-access-control)
205. Microsoft. *Restricted SharePoint Search*（文档）. [Microsoft Learn](https://learn.microsoft.com/en-us/sharepoint/restricted-sharepoint-search)

## RAG 评测与真实案例

206. Liang, X., et al. (2025). *SafeRAG: Benchmarking Security in Retrieval-Augmented Generation of Large Language Model*. [arXiv:2501.18636](https://arxiv.org/abs/2501.18636)
207. Yi, J., et al. (2023). *Benchmarking and Defending Against Indirect Prompt Injection Attacks on Large Language Models*（BIPIA）. [arXiv:2312.14197](https://arxiv.org/abs/2312.14197)
208. Fasolino. (2026). *In RAG We Trust? Measuring Robustness of Retrieval-Augmented Generation Under Post-Retrieval Context Tampering*. [arXiv:2609.09243](https://arxiv.org/abs/2609.09243)
209. Sadeghi, M., & Blachez, I. / NewsGuard. (2025). *A well-funded Moscow-based global 'news' network has infected Western artificial intelligence tools worldwide with Russian propaganda*. [NewsGuard Reality Check](https://www.newsguardrealitycheck.com/p/a-well-funded-moscow-based-global)
210. Microsoft Security. (2026). *Manipulating AI memory for profit: The rise of AI Recommendation Poisoning*. [Microsoft Security Blog](https://www.microsoft.com/en-us/security/blog/2026/02/10/ai-recommendation-poisoning/)
211. Evershed, N. / The Guardian. (2024). *ChatGPT search tool vulnerable to manipulation and deception, tests show*. [The Guardian](https://www.theguardian.com/technology/2024/dec/24/chatgpt-search-tool-vulnerable-to-manipulation-and-deception-tests-show)
212. Noma Security. (2025). *Hacking Google Gemini Enterprise with an Indirect Prompt Injection*（GeminiJack）. [Noma Labs](https://noma.security/noma-labs/geminijack)
213. PromptArmor. (2024). *Data Exfiltration from Slack AI via indirect prompt injection*. [PromptArmor](https://promptarmor.substack.com/p/data-exfiltration-from-slack-ai-via)
214. RoyChowdhury, A., et al. (2024). *ConfusedPilot: Confused Deputy Risks in RAG-based LLMs*. [arXiv:2408.04870](https://arxiv.org/abs/2408.04870)

## 智能体记忆投毒与防御

215. Dash, et al. (2026). *From Untrusted Input to Trusted Memory: A Systematic Study of Memory Poisoning Attacks in LLM Agents*（MPBench）. [arXiv:2606.04329](https://arxiv.org/abs/2606.04329)
216. Karunanidhi. (2026). *Utility Under Attack: Agent Memory Poisoning and the Limits of Content Screening and Provenance Ranking*. [arXiv:2608.21230](https://arxiv.org/abs/2608.21230)
217. Sharma. (2026). *SMSR: Certified Defence Against Runtime Memory Poisoning in Persistent LLM Agent Systems*. [arXiv:2606.12703](https://arxiv.org/abs/2606.12703)
218. Ouyang, & Hou. (2026). *MemLineage: Lineage-Guided Enforcement for LLM Agent Memory*. [arXiv:2605.14421](https://arxiv.org/abs/2605.14421)
219. Wei, et al. (2025). *A-MemGuard: A Proactive Defense Framework for LLM-Based Agent Memory*. [arXiv:2510.02373](https://arxiv.org/abs/2510.02373)
220. Rehberger, J. (2025). *Hacking Gemini's Memory with Prompt Injection and Delayed Tool Invocation*. [Embrace The Red](https://embracethered.com/blog/posts/2025/gemini-memory-persistence-prompt-injection/)

## 向量数据库多租户文档

221. Weaviate. *Multi-tenancy operations*（文档）. [Weaviate Docs](https://docs.weaviate.io/weaviate/manage-collections/multi-tenancy)
222. Milvus. *Implement Multi-tenancy*（文档）. [Milvus Docs](https://milvus.io/docs/multi_tenancy.md)
223. Qdrant. *Multitenancy*（文档）. [Qdrant Docs](https://qdrant.tech/documentation/guides/multiple-partitions/)
224. Pinecone. *Implement multitenancy*（文档）. [Pinecone Docs](https://docs.pinecone.io/guides/index-data/implement-multitenancy)
225. Nazary, F., Deldjoo, Y., & di Noia, T. (2025). *Poison-RAG: Adversarial Data Poisoning Attacks on Retrieval-Augmented Generation in Recommender Systems*. [arXiv:2501.11759](https://arxiv.org/abs/2501.11759)
226. Trellner. (2026). *Three sites made 215,128 "best software" pages for AI. Perplexity cites them*. [Trellner](https://trellner.com/reports/manufactured-sources-behind-ai-recommendations/)

## 智能体安全工具与方案文档

227. Open Policy Agent. *Open Policy Agent Documentation*. [OPA Docs](https://www.openpolicyagent.org/docs)
228. Open Policy Agent. (2025). *Note from Teemu, Tim, and Torin to the Open Policy Agent community*. [OPA Blog](https://www.openpolicyagent.org/blog/note-from-teemu-tim-and-torin-to-the-open-policy-agent-community-2dbbfe494371)
229. Cedar Policy. *Implementation of the Cedar Policy Language*（仓库）. [GitHub](https://github.com/cedar-policy/cedar)
230. AWS. *Policy in Amazon Bedrock AgentCore: Core concepts*（文档）. [AWS Docs](https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/policy-core-concepts.html)
231. Karanjkar, D. (2026). *Enforce least-privilege authorization in multi-agent AI chains using Cedar*. [AWS Security Blog](https://aws.amazon.com/blogs/security/enforce-least-privilege-authorization-in-multi-agent-ai-chains-using-cedar/)
232. OpenFGA. *AI Agent Authorization*（文档）. [OpenFGA Docs](https://openfga.dev/docs/use-cases/ai-agent-authorization)
233. NVIDIA. *NeMo Guardrails Library Developer Guide: Overview*（文档）. [NVIDIA Docs](https://docs.nvidia.com/nemo/guardrails/about-nemo-guardrails-library/overview)
234. Guardrails AI. *guardrails*（仓库 README，含停止托管远程推理的公告）. [GitHub](https://github.com/guardrails-ai/guardrails)
235. Meta. *LlamaFirewall*（PurpleLlama 仓库）. [GitHub](https://github.com/meta-llama/PurpleLlama/tree/main/LlamaFirewall)
236. Invariant Labs. *Invariant Guardrails*（仓库）. [GitHub](https://github.com/invariantlabs-ai/invariant)
237. Invariant Labs. *Invariant Explorer*（仓库 README，含托管版关停说明）. [GitHub](https://github.com/invariantlabs-ai/explorer)
238. Check Point. (2025). *Check Point to Acquire Lakera, Redefining Security for the AI era*. [Check Point Blog](https://blog.checkpoint.com/security/check-point-to-acquire-lakera-redefining-security-for-the-ai-era/)
239. Lakera / Check Point. *Guard API Endpoint*（Check Point AI Guardrails 文档）. [Lakera Docs](https://docs.lakera.ai/docs/api/guard)
240. AWS. *What are Automated Reasoning checks in Amazon Bedrock Guardrails?*（文档）. [AWS Docs](https://docs.aws.amazon.com/bedrock/latest/userguide/guardrails-automated-reasoning-checks.html)
241. Google Cloud. *Model Armor overview*（文档）. [Google Cloud Docs](https://docs.cloud.google.com/model-armor/overview)
242. LiteLLM. *LiteLLM AI Gateway (LLM Proxy)*（文档）. [LiteLLM Docs](https://docs.litellm.ai/docs/simple_proxy)
243. LiteLLM. *MCP Overview*（MCP 网关文档）. [LiteLLM Docs](https://docs.litellm.ai/docs/mcp)
244. Kong. *Kong AI Gateway*（文档）. [Kong Docs](https://developer.konghq.com/ai-gateway/)
245. Kong. *AI MCP Proxy plugin*（文档）. [Kong Docs](https://developer.konghq.com/plugins/ai-mcp-proxy/)
246. Cloudflare. *AI Gateway*（文档）. [Cloudflare Docs](https://developers.cloudflare.com/ai-gateway/)
247. Cloudflare. *AI Security for Apps*（原 Firewall for AI，文档）. [Cloudflare Docs](https://developers.cloudflare.com/waf/detections/ai-security-for-apps/)
248. Portkey. *AI Gateway*（仓库）. [GitHub](https://github.com/Portkey-AI/gateway)
249. Linux Foundation. (2025). *Linux Foundation Welcomes Agentgateway Project to Accelerate AI Agent Adoption While Maintaining Security, Observability, and Governance*. [Linux Foundation](https://www.linuxfoundation.org/press/linux-foundation-welcomes-agentgateway-project-to-accelerate-ai-agent-adoption-while-maintaining-security-observability-and-governance)
250. IBM. *ContextForge*（仓库）. [GitHub](https://github.com/IBM/mcp-context-forge)
251. Docker. *MCP Gateway*（文档）. [Docker Docs](https://docs.docker.com/ai/mcp-catalog-and-toolkit/mcp-gateway/)
252. Microsoft. *MCP Gateway*（文档）. [Microsoft](https://microsoft.github.io/mcp-gateway/)
253. Lasso Security. *MCP Gateway*（仓库）. [GitHub](https://github.com/lasso-security/mcp-gateway)
254. Snyk. *Snyk Agent Scan*（仓库 README）. [GitHub](https://github.com/snyk/agent-scan)
255. Model Context Protocol. *MCP Registry documentation*（含命名空间认证、审核政策与常见问题）. [GitHub](https://github.com/modelcontextprotocol/registry/tree/main/docs)
256. Anthropic. *sandbox-runtime*（仓库 README）. [GitHub](https://github.com/anthropics/sandbox-runtime)
257. OpenAI. *Sandbox*（Codex 文档）. [OpenAI](https://learn.chatgpt.com/docs/sandboxing)
258. Google. *What is gVisor?*（文档）. [gVisor](https://gvisor.dev/docs/)
259. Modal. *Sandboxes*（文档）. [Modal Docs](https://modal.com/docs/guide/sandboxes)
260. Firecracker. *Secure and fast microVMs for serverless computing*（仓库 README）. [GitHub](https://github.com/firecracker-microvm/firecracker)
261. E2B. *E2B Documentation*. [E2B Docs](https://docs.e2b.dev/)
262. E2B. *E2B runtime*（自托管运行时仓库 README）. [GitHub](https://github.com/e2b-dev/runtime)
263. microsandbox. *microsandbox*（仓库 README）. [GitHub](https://github.com/superradcompany/microsandbox)
264. Daytona. *daytona*（仓库 README，含停止维护声明）. [GitHub](https://github.com/daytonaio/daytona)
265. SPIFFE. *SPIFFE Overview*（文档）. [SPIFFE](https://spiffe.io/docs/latest/spiffe-about/overview/)
266. Keycloak. *Configuring and using token exchange*（文档）. [Keycloak Docs](https://www.keycloak.org/securing-apps/token-exchange)
267. Auth0. (2025). *Auth0 for AI Agents is Now Generally Available (GA)*. [Auth0 Blog](https://auth0.com/blog/auth0-for-ai-agents-generally-available/)
268. Okta. *Cross App Access (XAA)*（概念文档）. [Okta Developer](https://developer.okta.com/docs/concepts/xaa/)
269. Okta. (2026). *Okta brings first-class identity to AI agents with Agent SSO*. [Okta Newsroom](https://www.okta.com/newsroom/press-releases/okta-brings-first-class-identity-to-ai-agents-with-agent-sso/)
270. HashiCorp. *Vault + agentic AI*（文档，含 Agentic IAM）. [HashiCorp Developer](https://developer.hashicorp.com/vault/ai)
271. HashiCorp. *What is the Vault MCP server?*（文档，测试版）. [HashiCorp Developer](https://developer.hashicorp.com/vault/ai/mcp-server)
272. 1Password. (2026). *1Password and Anthropic Bring Secure Credential Access to Claude*（新闻稿）. [1Password](https://1password.com/press/2026/july/1password-for-claude)
273. Langfuse. *Langfuse Documentation*. [Langfuse Docs](https://langfuse.com/docs)
274. Langfuse. (2026). *Langfuse joins ClickHouse*. [Langfuse Blog](https://langfuse.com/blog/joining-clickhouse)
275. Arize. *Phoenix: AI Observability & Evaluation*（仓库，含 Elastic License 2.0 许可文件）. [GitHub](https://github.com/Arize-ai/phoenix)
276. LangChain. *Self-hosted LangSmith*（文档）. [LangSmith Docs](https://docs.langchain.com/langsmith/self-hosted)
277. Promptfoo. *Promptfoo Documentation: Intro*. [Promptfoo Docs](https://www.promptfoo.dev/docs/intro/)
278. Confident AI. *DeepTeam: Quick Introduction*（文档）. [DeepTeam Docs](https://www.trydeepteam.com/docs/getting-started)
279. Microsoft. *AI Red Teaming Agent*（Microsoft Foundry 文档）. [Microsoft Learn](https://learn.microsoft.com/en-us/azure/foundry/concepts/ai-red-teaming-agent)
280. Snyk. (2025). *Snyk Acquires Invariant Labs to Accelerate Agentic AI Security Innovation*. [Snyk](https://snyk.io/news/snyk-acquires-invariant-labs-to-accelerate-agentic-ai-security-innovation/)

## RAG 章补充来源

281. OWASP. *LLM Prompt Injection Prevention Cheat Sheet*. [OWASP Cheat Sheet Series](https://cheatsheetseries.owasp.org/cheatsheets/LLM_Prompt_Injection_Prevention_Cheat_Sheet.html)
282. AWS. *Controlling user access to documents with tokens*（Amazon Kendra 文档）. [AWS Docs](https://docs.aws.amazon.com/kendra/latest/dg/create-index-access-control.html)
283. Nguyen, & Husain. (2026). *An Experimental Evaluation of Multimodal Prompt Injection Attacks on Agentic AI Frameworks*. [arXiv:2609.09404](https://arxiv.org/abs/2609.09404)
284. Thornton. (2026). *Semantic Chameleon: Corpus-Dependent Poisoning Attacks and Defenses in RAG Systems*. [arXiv:2603.18034](https://arxiv.org/abs/2603.18034)
285. Meta. (2025). *Llama Guard 4 Model Card*（meta-llama/Llama-Guard-4-12B）. [Hugging Face](https://huggingface.co/meta-llama/Llama-Guard-4-12B)
286. Padhi, I., Nagireddy, M., et al. (2024). *Granite Guardian*. [arXiv:2412.07724](https://arxiv.org/abs/2412.07724)
287. IBM. (2025). *Granite Guardian 3.3 8B Model Card*（ibm-granite/granite-guardian-3.3-8b）. [Hugging Face](https://huggingface.co/ibm-granite/granite-guardian-3.3-8b)
288. Zeng, W., Liu, Y., et al. (2024). *ShieldGemma: Generative AI Content Moderation Based on Gemma*. [arXiv:2407.21772](https://arxiv.org/abs/2407.21772)
289. Google. *ShieldGemma*（安全分类器文档）. [Google AI for Developers](https://ai.google.dev/responsible/docs/safeguards/shieldgemma)
290. Zeng, W., Kurniawan, D., et al. (2025). *ShieldGemma 2: Robust and Tractable Image Content Moderation*. [arXiv:2504.01081](https://arxiv.org/abs/2504.01081)
291. Han, S., Rao, K., et al. (2024). *WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMs*. [arXiv:2406.18495](https://arxiv.org/abs/2406.18495)
292. Dathathri, S., See, A., Ghaisas, S., et al. (2024). *Scalable watermarking for identifying large language model outputs*. Nature, 634, 818–823. [DOI](https://doi.org/10.1038/s41586-024-08025-4)
293. Google DeepMind. *SynthID Text*（参考实现仓库，仅供研究复现）. [GitHub](https://github.com/google-deepmind/synthid-text)
294. Google. *SynthID: Tools for watermarking and detecting LLM-generated Text*. [Google AI for Developers](https://ai.google.dev/responsible/docs/safeguards/synthid)
295. OWASP CycloneDX. *Authoritative Guide to AI/ML-BOM*（First Edition）. [CycloneDX](https://cyclonedx.org/guides/OWASP_CycloneDX-Authoritative-Guide-to-AI-ML-BOM-en.pdf)
296. OWASP CycloneDX. (2023). *CycloneDX v1.5 Released*. [CycloneDX](https://cyclonedx.org/news/cyclonedx-v1.5-released/)
297. Sigstore. *model-transparency: Supply chain security for ML*（model-signing）. [GitHub](https://github.com/sigstore/model-transparency)
298. ml-guard. *ML Guard: Security and compliance scanning for machine learning pipelines*. [GitHub](https://github.com/ml-guard/ml-guard)

## 智能体网络层身份与通信

299. A2A Project. (2025). *A2A v0.3.0*（发布说明：加入 mTLS 安全方案，卡片路径改为 agent-card.json）. [GitHub](https://github.com/a2aproject/A2A/releases/tag/v0.3.0)
300. A2A Project. *Enterprise Features*（A2A 文档）. [A2A Protocol](https://a2a-protocol.org/latest/topics/enterprise-ready/)
301. SPIRE. *SPIRE Server Configuration Reference*（含 SVID 默认有效期）. [GitHub](https://github.com/spiffe/spire/blob/main/doc/spire_server.md)
302. IETF WIMSE Working Group. (2026). *Workload Identity in a Multi System Environment (WIMSE) Architecture*（draft-ietf-wimse-arch-08，Internet-Draft）. [IETF Datatracker](https://datatracker.ietf.org/doc/draft-ietf-wimse-arch/)
303. Istio. *Security*（概念文档，含对等认证模式与证书轮换）. [Istio Docs](https://istio.io/latest/docs/concepts/security/)
304. Istio. *pilot-discovery*（命令参考，含工作负载证书默认有效期）. [Istio Docs](https://istio.io/latest/docs/reference/commands/pilot-discovery/)
305. Linkerd. *Automatic mTLS*（文档）. [Linkerd Docs](https://linkerd.io/docs/features/automatic-mtls/)
306. Google Cloud. *Agent Identity overview*（文档）. [Google Cloud Docs](https://docs.cloud.google.com/iam/docs/agent-identity-overview)
307. Google Cloud. *IAM release notes*（Agent Identity 正式发布记录）. [Google Cloud Docs](https://docs.cloud.google.com/iam/docs/release-notes)
308. Microsoft. *Agent identities in Microsoft Entra Agent ID*（文档）. [Microsoft Learn](https://learn.microsoft.com/en-us/entra/agent-id/agent-identities)
309. AWS. *Get a workload access token*（Amazon Bedrock AgentCore Identity 文档）. [AWS Docs](https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/get-workload-access-token.html)
310. OWASP GenAI Security Project. (2025). *Securing Agentic Applications Guide 1.0*. [OWASP](https://genai.owasp.org/resource/securing-agentic-applications-guide-1-0/)
311. Let's Encrypt. (2026). *6-day and IP Address Certificates are Generally Available*. [Let's Encrypt](https://letsencrypt.org/2026/01/15/6day-and-ip-general-availability)
312. CA/Browser Forum. (2025). *Ballot SC081v3: Introduce Schedule of Reducing Validity and Data Reuse Periods*. [CA/Browser Forum](https://cabforum.org/2025/04/11/ballot-sc081v3-introduce-schedule-of-reducing-validity-and-data-reuse-periods/)
313. Model Context Protocol. (2026). *Sampling*（规范 2026-07-28 修订版，该功能已弃用）. [MCP Specification](https://modelcontextprotocol.io/specification/2026-07-28/client/sampling)
314. Model Context Protocol. (2026). *Elicitation*（规范 2026-07-28 修订版）. [MCP Specification](https://modelcontextprotocol.io/specification/2026-07-28/client/elicitation)

## 编码智能体的奖励投机

315. Von Arx, S., Chan, L., & Barnes, B. (METR). (2025). *Recent Frontier Models Are Reward Hacking*. [METR](https://metr.org/blog/2025-06-05-recent-reward-hacking/)
316. Anthropic. (2025). *System Card: Claude Opus 4 & Claude Sonnet 4*. [Anthropic](https://www-cdn.anthropic.com/4263b940cabb546aa0e3283f35b686f4f3b2ff47.pdf)
317. Anthropic. (2025). *Claude 3.7 Sonnet System Card*. [Anthropic](https://assets.anthropic.com/m/785e231869ea8b3b/original/claude-3-7-sonnet-system-card.pdf)
318. OpenAI. (2025). *Detecting misbehavior in frontier reasoning models*（论文：Baker et al., *Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation*, [arXiv:2503.11926](https://arxiv.org/abs/2503.11926)）. [OpenAI](https://openai.com/index/chain-of-thought-monitoring/)
319. MacDiarmid, M., Wright, B., Uesato, J., et al. (Anthropic). (2025). *Natural Emergent Misalignment from Reward Hacking in Production RL*. [arXiv:2511.18397](https://arxiv.org/abs/2511.18397)
320. Zhong, Z., Raghunathan, A., & Carlini, N. (2025). *ImpossibleBench: Measuring LLMs' Propensity of Exploiting Test Cases*. [arXiv:2510.20270](https://arxiv.org/abs/2510.20270)
321. SWE-bench. (2025). *Repo State Loopholes During Agentic Evaluation*（Issue #465）. [GitHub](https://github.com/SWE-bench/SWE-bench/issues/465)
322. Anthropic. *Best practices for Claude Code*. [Claude Code Docs](https://code.claude.com/docs/en/best-practices)

## 智能体支付协议

323. Google. (2026). *AP2 Security and Privacy Considerations*. [GitHub](https://github.com/google-agentic-commerce/AP2/blob/main/docs/ap2/security_and_privacy_considerations.md)
324. Google. (2026). *Agent Payments Protocol (AP2) Specification*（0.2 版）. [GitHub](https://github.com/google-agentic-commerce/AP2/blob/main/docs/ap2/specification.md)
325. Google. (2026). *Google donates Agent Payments Protocol to FIDO Alliance*. [The Keyword](https://blog.google/products-and-platforms/platforms/google-pay/agent-payments-protocol-fido-alliance/)
326. OpenAI & Stripe. *Agentic Commerce Protocol: Delegate Payment RFC*. [GitHub](https://github.com/agentic-commerce-protocol/agentic-commerce-protocol/blob/main/rfcs/rfc.delegate_payment.md)
327. OpenAI. *Delegated Payment Spec*. [OpenAI Developers](https://developers.openai.com/commerce/specs/payment)
328. x402 Foundation. *x402 Protocol Specification v2*. [GitHub](https://github.com/x402-foundation/x402/blob/main/specs/x402-specification-v2.md)
329. x402 Foundation. *x402 FAQ*. [GitHub](https://github.com/x402-foundation/x402/blob/main/docs/faq.md)
330. Linux Foundation. (2026). *Linux Foundation Announces Operational Launch of x402 Foundation*. [x402.org](https://x402.org/linux-foundation-announces-operational-launch-of-x402-foundation-to-standardize-internet-native-payments-for-ai-agents-and-applications/)
331. x402 Foundation. *Scheme: exact on EVM*. [GitHub](https://github.com/x402-foundation/x402/blob/main/specs/schemes/exact/scheme_exact_evm.md)

## 代码编排的工具调用

332. Anthropic. *Programmatic tool calling*（文档）. [Claude Docs](https://platform.claude.com/docs/en/agents-and-tools/tool-use/programmatic-tool-calling)
333. Anthropic. (2025). *Code execution with MCP: Building more efficient agents*. [Anthropic Engineering](https://www.anthropic.com/engineering/code-execution-with-mcp)

## MCP 规范变更

334. Model Context Protocol. (2026). *Key Changes*（规范 2026-07-28 修订版变更记录，含多轮往返请求）. [MCP Specification](https://modelcontextprotocol.io/specification/2026-07-28/changelog)

## 治理法规与框架修订

335. European Union. (2026). *Regulation (EU) 2026/1744 amending Regulation (EU) 2024/1689*（含 Article 111(4)、Article 113 时间线修订）. [EUR-Lex](https://eur-lex.europa.eu/eli/reg/2026/1744/oj/eng)
336. Board of Governors of the Federal Reserve System. (2026). *SR 26-2: Revised Guidance on Model Risk Management*（替代 SR 11-7；附件脚注 3 排除生成式 AI 与智能体 AI）. [Federal Reserve](https://www.federalreserve.gov/supervisionreg/srletters/SR2602.htm)；[正式附件](https://www.federalreserve.gov/supervisionreg/srletters/SR2602a1.pdf)
337. Anthropic. (2026). *Responsible Scaling Policy, Version 3.4*（Frontier Safety Roadmap、Risk Reports 与 Appendix A 的条件承诺）. [Anthropic PDF](https://www-cdn.anthropic.com/files/4zrzovbb/website/0bacdc8440ea96e62a8766d99ebe1d4eea6d5f3a.pdf)

## 动态污点分析与指令—数据分离

338. Newsome, J., & Song, D. (2005). *Dynamic Taint Analysis for Automatic Detection, Analysis, and Signature Generation of Exploits on Commodity Software*. NDSS 2005. [PDF](https://valgrind.org/docs/newsome2005.pdf)
339. Zverev, E., Abdelnabi, S., Tabesh, S., Fritz, M., et al. (2024). *Can LLMs Separate Instructions From Data? And What Do We Even Mean By That?*（SEP 基准）. [arXiv:2403.06833](https://arxiv.org/abs/2403.06833)
340. Wu, T., Zhang, S., Song, K., Xu, S., et al. (2024). *Instructional Segment Embedding: Improving LLM Safety with Instruction Hierarchy*（ISE）. [arXiv:2410.09102](https://arxiv.org/abs/2410.09102)
341. Zverev, E., Kortukov, E., Panfilov, A., Volkova, A., et al. (2025). *ASIDE: Architectural Separation of Instructions and Data in Language Models*. [arXiv:2503.10566](https://arxiv.org/abs/2503.10566)
342. Chismon, D. / UK NCSC. (2025). *Prompt injection is not SQL injection (it may be worse)*. [NCSC](https://www.ncsc.gov.uk/blog-post/prompt-injection-is-not-sql-injection)

## 完整性模型

343. Biba, K. J. (1977). *Integrity Considerations for Secure Computer Systems*. ESD-TR-76-372, USAF Electronic Systems Division, April 1977（MITRE MTR-3153 初版为 1975 年 6 月）.
344. Clark, D. D., & Wilson, D. R. (1987). *A Comparison of Commercial and Military Computer Security Policies*. IEEE Symposium on Security and Privacy 1987；全文收入 NIST SP 500-160 附录 A1. [NIST](https://nvlpubs.nist.gov/nistpubs/Legacy/SP/nistspecialpublication500-160.pdf)
345. Fraser, T. (2001). *LOMAC: MAC You Can Live With*. USENIX Annual Technical Conference, FREENIX Track（LOMAC 最早发表于 IEEE S\&P 2000）. [USENIX](https://static.usenix.org/event/usenix01/freenix01/full_papers/fraser/fraser.pdf)

## 智能体安全的三难表述

346. Raghavan, B., & Schneier, B. (2025). *Agentic AI's OODA Loop Problem*. IEEE Security & Privacy, September/October 2025. [Schneier on Security](https://www.schneier.com/essays/archives/2025/10/agentic-ais-ooda-loop-problem.html)
347. Neray, G. / Oso. (2026). *The CAP Theorem for Agents*. [Oso Blog](https://www.osohq.com/post/the-cap-theorem-for-agents)
348. Bhatt, M., Munshi, S., et al. (2026). *The Defense Trilemma: Why Prompt Injection Defense Wrappers Fail?*. [arXiv:2604.06436](https://arxiv.org/abs/2604.06436)
349. Abdelnabi, S., & Bagdasarian, E. (2026). *AI Agents May Always Fall for Prompt Injections*. [arXiv:2605.17634](https://arxiv.org/abs/2605.17634)
350. Black Flag. (2025). *"Pick Two" AI Trilemma: Generality, Agency, Alignment*. [LessWrong](https://www.lesswrong.com/posts/LrwXC2HZpB494ASZS/pick-two-ai-trilemma-generality-agency-alignment)

## 指令层级训练

351. Wallace, E., Xiao, K., Leike, R., Weng, L., Heidecke, J., & Beutel, A. (2024). *The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions*. [arXiv:2404.13208](https://arxiv.org/abs/2404.13208)

***

*参考文献会随时间变化，后续版本将持续更新。*
