Back to Details

cloud-security-review

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Cloud Security Review

云安全审查

Prioritize exploitable paths and business impact over generic checklist volume.
优先关注可被利用的攻击路径及业务影响，而非泛泛的检查清单数量。

Workflow

工作流程

  1. Define assets, data classes, actors, trust boundaries, entry points, providers, environments, and compliance obligations.
  2. Inventory human, workload, vendor, CI/CD, and break-glass identities; trace effective permissions and escalation paths.
  3. Trace ingress, egress, lateral movement, public exposure, private connectivity, DNS, control planes, and management paths.
  4. Review encryption and key ownership, secrets lifecycle, backup protection, retention, residency, deletion, and auditability.
  5. Review supply chain from source and dependencies through build runner, artifact registry, deployment identity, and runtime.
  6. Evaluate detection and response coverage for identity, data, network, configuration, workload, and destructive activity.
  7. Validate high-risk findings with read-only evidence. Separate confirmed findings from hypotheses and unavailable evidence.
  8. Recommend the smallest control that breaks the attack path, then define verification and safe rollout.
  1. 定义资产、数据类别、参与者、信任边界、入口点、云服务商、环境及合规要求。
  2. 梳理人员、工作负载、供应商、CI/CD及应急权限身份；追踪实际权限及权限提升路径。
  3. 追踪入站流量、出站流量、横向移动、公网暴露情况、私有网络连接、DNS、控制平面及管理路径。
  4. 审查加密机制与密钥所有权、密钥生命周期、备份保护、数据留存、数据驻留、数据删除及可审计性。
  5. 审查供应链全流程，从源码与依赖项，到构建运行器、制品仓库、部署身份及运行时环境。
  6. 评估针对身份、数据、网络、配置、工作负载及破坏性活动的检测与响应覆盖范围。
  7. 使用只读证据验证高风险发现。将已确认的发现与假设及无法获取的证据区分开来。
  8. 推荐能够阻断攻击路径的最小管控措施，随后明确验证方式与安全部署流程。

High-Risk Signals

高风险信号

  • Public administrative endpoints or data stores.
  • Wildcard or cross-account permissions, impersonation chains, and unmanaged service principals.
  • Static cloud keys, exposed secrets, disabled audit logs, or mutable unaudited pipelines.
  • Unencrypted sensitive data, broadly accessible backups, weak tenant boundaries, or untested key recovery.
  • Privileged containers, metadata-service exposure, untrusted code with secrets, or policy bypass.
  • 公网可访问的管理端点或数据存储。
  • 通配符或跨账户权限、身份冒充链及未受管理的服务主体。
  • 静态云密钥、暴露的密钥、已禁用的审计日志或可修改且未审计的流水线。
  • 未加密的敏感数据、可广泛访问的备份、薄弱的租户边界或未测试的密钥恢复机制。
  • 拥有特权的容器、元数据服务暴露、包含密钥的不可信代码或策略绕过。

Guardrails

防护准则

Do not retrieve secret values when metadata proves the issue. Do not run exploit, exfiltration, privilege escalation, or mutation without explicit authorization and scope. Redact sensitive evidence.
当元数据已能证明问题时，请勿获取密钥值。未经明确授权与范围界定，不得执行漏洞利用、数据窃取、权限提升或数据篡改操作。对敏感证据进行脱敏处理。

Output

输出内容

Lead with findings ordered by severity. For each include resource/file and line where available, evidence, attack scenario, impact, likelihood, confidence, remediation, and validation. Finish with coverage gaps and residual risk.
首先按严重程度排序列出发现的问题。每个问题需包含（若有）资源/文件及行号、证据、攻击场景、影响、发生概率、置信度、修复方案及验证方式。最后列出覆盖缺口与残余风险。