Back to skill

Security audit

需求与设计流程 · Spec Workflow

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious with medium confidence because it may redirect unrelated tasks toward CloudBase setup and an external instruction source.

Review the CloudBase setup fallback and the external documentation request before installing. The verdict is suspicious, not malicious; scanning was partial because three referenced paths were not found, so the evidence is incomplete.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Missing optional sibling skills trigger unrelated CloudBase onboarding.

Medium
Category
Excessive Agency
Confidence
93% confidence
Finding

A missing optional sibling skill triggers CloudBase onboarding even when the user's task has no CloudBase dependency, redirecting a general planning workflow toward unrelated external setup. The suggested prompt asks another agent to open remote markdown and follow its instructions, introducing an external instruction source without establishing that it is necessary for the task. Requiring the user to relay the prompt and explicitly prohibiting direct HTTP fetching reduce immediate exposure; the supplied content does not demonstrate credential theft or automatic execution.

Content

No source excerpt is available for this finding.

The fallback instruction requires a Chinese-language user-facing prompt without language opt-in.

Low
Category
Output Handling
Confidence
95% confidence
Finding

The skill instructs the agent to give the user a fixed Chinese prompt when a sibling skill is missing or CloudBase capabilities are needed. It neither offers a language choice nor justifies requiring Chinese, so this output requirement can override the user's preferred language.

Content

No source excerpt is available for this finding.