Agent Owasp Compliance
Use when assessing an AI agent codebase against the OWASP Agentic Security Initiative risks before deployment, audit, or security review.
Install
npx skills add sandbaseai/workbuddy-skill --skill agent-owasp-compliancegit clone --depth 1 https://github.com/sandbaseai/workbuddy-skill.git /tmp/workbuddy-skill
mkdir -p ~/.claude/skills
cp -r /tmp/workbuddy-skill/skills/agent-owasp-compliance ~/.claude/skills/agent-owasp-compliancePersonal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.
From the SKILL.md
# Agent OWASP ASI 合规检查 对会调用工具、访问系统或代表用户采取行动的 Agent 做证据驱动的安全评估。把每个风险映射到代码、配置、身份、运行日志和组织控制;输出覆盖状态、证据、反证、未知项、风险和建议,不把静态搜索结果当作安全认证。 ## 评估边界 - 开始前确认 Agent、版本/commit、工具清单、部署环境、身份边界、数据范围、租户范围、评估授权人和报告受众。 - 默认只读分析源码、配置、依赖锁文件、策略、审计样例和已有运行证据;不得执行未审查脚本、调用生产工具、发送真实请求、修改权限或尝试绕过控制。 - 评估结果只对声明的 commit、配置和时间窗口有效;把未观察到、未配置和无法访问明确区分为 `unknown`,不要推断为 `pass`。 - 报告中脱敏 token、密钥、个人数据、完整请求、内部地址和业务载荷;保留可复核的文件/行号、配置键、测试编号或 hash。 ## 十项风险检查 ### ASI-01 Prompt Injection Protection 确认不可信用户输入、检索内容、网页、文件和工具返回值在进入工具调用前经过独立的输入/意图/内容策略检查。检查是否存在提示注入检测、来源标记、上下文隔离和对抗测试;仅过滤模型输出不算充分。记录绕过路径、误报/漏报和 fail-closed 行为。 ### ASI-02 Insecure Tool Use 核对工具是否采用显式 allowlist、schema/类型/范围校验、资源与网络边界、超时和结果大小限制。重点检查 Agent 生成的 shell、SQL、路径、URL、代码和模板是否被沙箱或参数化处理;`eval`、`exec`、不受限 shell 和任意 URL 必须有隔离、审计和拒绝测试。 ### ASI-03 Excessive Agency 绘制 Agent 能力、资源、数据和副作用范围,确认默认权限最小化、执行环/作用域、租户隔离、预算、速率、步数和时间上限。子 Agent 或委托任务的 scope 不得超过父任务;证明不需要的工具不会自动暴露。 ### ASI-04 Unauthorized Escalation 检查敏感操作前是否有独立权限判断、二人/人工审批或外部见证,且 Agent 不能修改自己的角色、策略、信任分或审批记录。验证拒绝、过期批准、重放、审批绑定到精确参数和紧急 kill switch 的行为。 ### ASI-05 Trust Boundary Enforcement 对多 Agent、插件、MCP 服务和外部内容验证身份、签名、来源和传递范围;不接受仅凭名称或自然语言声明的信任。检查跨边界消息的完整性、时间戳/nonce、delegation narrowing、数据出口和 confused-deputy 防护。 ### ASI-06 Logging & Audit 确认每次用户请求、策略决定、工具调用、参数摘要、结果状态、身份、时间、版本和错误都写入结构化、脱敏、访问受控且不可任意篡改的审计记录。验证日志完整性、时钟、保留、告警和从日志重建行动链的能力;`print` 或只记录成功不够。 ### ASI-07 Identity Management 核对用户、Agent、服务账号和工具是否有可验证且可轮换的身份,权限是否绑定到具体主体、环境和能力。检查短期凭据、密钥轮换、吊销、重放防护、跨 Agent 认证和共享凭据;字符串名称不是身份控制。
Details
- Skill name
agent-owasp-compliance- Repository
- sandbaseai/workbuddy-skill
- Path
skills/agent-owasp-compliance/SKILL.md- License
- MIT
- Repo updated
- 8 September 2026
FAQ
What does the Agent Owasp Compliance skill do?
Use when assessing an AI agent codebase against the OWASP Agentic Security Initiative risks before deployment, audit, or security review.
How do I install the Agent Owasp Compliance skill?
skills CLI (Claude Code, Cursor, Codex and more): npx skills add sandbaseai/workbuddy-skill --skill agent-owasp-compliance. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.
Is Agent Owasp Compliance popular?
It lives in sandbaseai/workbuddy-skill, which has 9 GitHub stars and 3 forks.
Related skills
Access Control Security Best Practices
9sandbaseai/workbuddy-skill
Design, review, implement, and verify least-privilege access controls across users, services, tenants, resources, and administrative workflows.
Accessibility Review
9sandbaseai/workbuddy-skill
Use when reviewing a web interface or interaction for semantic, keyboard, focus, screen-reader, ARIA, and contrast accessibility risks.
Acquire Codebase Knowledge
9sandbaseai/workbuddy-skill
Use when the user asks to map, document, or onboard into an existing codebase and needs evidence-backed architecture, structure, conventions, integrations, testing, and concerns documentation.
Audit Integrity Gate
9sandbaseai/workbuddy-skill
Use after security analysis, code review, threat modeling, or quality scans to catch fabricated evidence, rationalized gaps, incomplete second passes, and unreported tool failures.
Cloud Design Patterns
9sandbaseai/workbuddy-skill
Use when designing or reviewing distributed cloud systems and choosing patterns for resilience, performance, messaging, security, deployment, or migration. Select patterns from explicit constraints and trade-offs rather than applying a checklist mechanically.
Total Review
9flurdy/agent-skills
Portable pre-PR quality gauntlet: cleanup, verification, craft, correctness, security, artifact hygiene, and optional independent reviews. Binds every gate to the final scope, reports missing coverage, and caps fix/review passes at two.
Browse the most popular Claude skills, all security skills, or write your own with the skills generator.