Back to skill

Security audit

F.A.C.E.T. Framework (心智模型:F.A.C.E.T. 框架)

Security checks for vulnerabilities and agentic risk

Overview

This skill is a markdown-only mental-model analysis template with limited, disclosed personalization from USER.md and no executable behavior.

Install this only if you are comfortable with the skill reading USER.md for personalization. Keep sensitive details out of USER.md if you do not want them reflected in generated analysis, and review any metadata before another tool stores it.

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
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Hidden Instructions

High
Category
Prompt Injection
Content
2. `### ⚡ Strategic Question` — one sharp, actionable question

### Section 2: Knowledge Base Metadata (for storage)
Structured metadata block at the end, fenced with `<!-- KB_META_START -->` and `<!-- KB_META_END -->`:

```yaml
<!-- KB_META_START -->
Confidence
70% confidence
Finding
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Natural-Language Policy Violations

Medium
Confidence
95% confidence
Finding
The skill mandates English/Chinese-specific output length constraints without asking for the user's preferred language or locale. While not directly a security exploit, this can override user expectations and cause unintended disclosure or mishandling of content in multilingual settings, especially if the user did not request those languages.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The skill explicitly instructs itself to read USER.md for background, interests, and current challenges, but the top-level description does not clearly warn users that personal context may be accessed and used for personalization. This creates a transparency and privacy-consent issue because sensitive profile data could be incorporated into outputs without the user's informed awareness.

Static analysis

No suspicious patterns detected.