First-Principles-Deconstructor

• Outcome: Pass
• Latest audit: May 29, 2026, 8:21 PM
• Version:
  1.0.1

Overview

This skill is a product and technology audit helper with limited disclosed handoff-file use and no executable code.

Before installing, understand that this skill may use prior agent output from ~/.openclaw/swarm_tmp/expert_output.json and save an audit report under ~/.openclaw/swarm_tmp/audit_report.json. Clear that handoff file if you do not want earlier task content to influence a new audit.

SkillSpector

By NVIDIA

Vulnerability Patterns

• 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
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The usage section defines very broad trigger phrases like 'research' and 'compare' without any scoping, namespace, or invocation constraints. In agent systems, underspecified triggers can cause accidental activation on ordinary user requests, leading to unintended behavior, misrouting, or the application of this skill in contexts where its aggressive evaluative style is not appropriate.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill explicitly instructs the agent to read from and write to local filesystem paths under ~/.openclaw/swarm_tmp without any user notice, consent, or scope limitation. Even though the paths appear application-specific, this creates an unauthorized local data access and persistence channel that can expose prior task artifacts, enable unintended cross-task data flow, and leave sensitive audit outputs on disk.

VirusTotal

66/66 vendors flagged this skill as clean.