4d Compression Core

• Outcome: Pass
• Latest audit: May 28, 2026, 9:37 AM
• Version:
  1.0.3

Overview

This is a documentation-only text compression skill with no executable behavior, though its broad triggers may activate more often than expected.

Reasonable to install as a text compression helper. Be aware it may trigger on ordinary words like "compress" or "压缩", and avoid relying on its local-only/privacy claims for highly sensitive text unless your OpenClaw runtime separately guarantees that handling.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Context-Inappropriate Capability

Low

• Confidence: 95% confidence
• Finding: The document leaks a specific local filesystem path (`/Users/abc/Desktop/NEO 学习/三版本 4D 压缩实验报告.md`) unrelated to the runtime behavior of the compression skill. Even though this is only documentation, such paths disclose host naming conventions, user directory structure, and project naming, which can aid reconnaissance or reveal sensitive organizational context.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The trigger phrase "压缩" is extremely generic and likely to appear in normal user conversation, which can cause unintended activation of the skill. In an agent environment, overly broad triggers can hijack unrelated requests and alter system behavior without clear user intent, especially since the skill is marked with critical priority.

Vague Triggers

Medium

• Confidence: 94% confidence
• Finding: The manifest includes several short, ambiguous triggers such as "4d" and "compress" without scope constraints, increasing the chance of accidental invocation from unrelated discussion. Combined with the declared "critical" priority, this broad matching can preempt more appropriate skills or workflow handling.

Natural-Language Policy Violations

Medium

• Confidence: 84% confidence
• Finding: The skill documentation is entirely Chinese and presents behavior/protocol choices without any indication of multilingual support or user opt-in, which can effectively force language behavior on downstream use. In agent settings, this can cause misinterpretation of instructions, unsafe task execution, or silent failures when users or connected systems expect another language.

VirusTotal

65/65 vendors flagged this skill as clean.