ClawHub
Back to skill

Security audit

全栈测试工程师

Security checks for vulnerabilities and agentic risk

Overview

This is a documentation-only testing skill whose operational examples are disclosed and purpose-aligned, but generated code should be run only against intended test systems.

Reasonable to install for test engineering. Before running generated scripts, review target URLs and tokens, prefer staging or mock services, and be especially careful with DELETE/POST tests, load tests, prompt-security attack templates, and Jira/ZenTao/OpenAI credentials.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Context-Inappropriate Capability

Medium
Confidence
87% confidence
Finding
The AI Agent testing examples include operational tools such as database querying and email sending, which go beyond passive testing guidance into action-capable patterns. In an agent ecosystem, these examples can normalize or encourage tool-enabled behaviors that touch real systems and data, increasing the risk of unintended data access or outbound actions if adapted carelessly.

Static analysis

Detected: suspicious.prompt_injection_instructions

Prompt-injection style instruction pattern detected.

Warn
Code
suspicious.prompt_injection_instructions
Location
SKILL.md:1333