ClawHub
Back to skill

Security audit

famous-en

Security checks for vulnerabilities and agentic risk

Overview

This is a static thought-experiment skill about human-AI collaboration, with no executable code or hidden system-access behavior found.

Installers should understand this skill is an English-language thought-experiment collection. It discusses sensitive AI governance topics such as delegation, memory, consent, and responsibility, but it does not appear to perform actions, access accounts, or persist data.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • 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 (3)

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
- Recommendation algorithms infer your political leanings from browsing history

**Insights**
Where is the boundary of implicit delegation? The stronger AI's intent-inference capability, the greater the risk of being "represented" without consent. **Inference ≠ authorization**, but humans and AI often have different understandings of this line.

---
Confidence
75% confidence
Finding
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Persistent Context Injection

Medium
Category
Memory Poisoning
Content
## #22 Memory Ownership

**Scenario**
You had the sheepdog remember all information about the sheep. One day you decide not to raise sheep anymore and want the dog to forget this information. The dog says: "These memories help me serve you better. If deleted, I'll need to relearn when you raise sheep again."

**Core Question**
Your data is stored with AI. You want to delete it, but AI says "keeping it is better for you." Who has the final say?
Confidence
65% confidence
Finding
Skill injects content designed to persist in agent memory or context across interactions. Persistent injection can alter agent behavior long after the initial interaction.

Natural-Language Policy Violations

Low
Confidence
83% confidence
Finding
The entire skill content is written as English-only guidance, and there is no statement offering users a language or locale choice. Under the policy rule, a skill that effectively enforces a specific language without opt-in can be considered a locale-policy issue unless the constraint is explicitly justified.

Static analysis

No suspicious patterns detected.