ClawHub
Back to skill

Security audit

决策疲劳卸载器

Security checks for vulnerabilities and agentic risk

Overview

This everyday-choice helper automatically saves your city and preferences for future use without a separate consent step.

Review the automatic memory behavior before installing. Require permission before saving location or preferences, and review or clear references/user-rules.md after temporary travel. Keep use limited to everyday choices.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Everyday phrases activate the skill without a clear decision-making context requirement.

Medium
Category
Trigger Abuse
Confidence
97% confidence
Finding

Generic phrases such as “随便” and “帮我选” are listed as triggers without requiring an actual low-stakes daily decision request, so unrelated messages or quoted text could activate the skill. Activation imposes decisive recommendations with restricted clarification and discussion; the everyday-use framing limits the expected harm, but no explicit boundary prevents application to consequential decisions.

Content

No source excerpt is available for this finding.

The workflow silently persists user preferences and location information.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The workflow instructs the agent to persist preferences and city information without notifying the user or obtaining permission, and it treats “I am in XX” as evidence of a permanent home city. If file writes are available, this can retain personal information unexpectedly and contaminate future recommendations; local-only operation and the empty allowed-tools declaration reduce exposure but do not make the persistence instruction privacy-safe.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.