PMP-Agentclaw

• Outcome: Pass
• Latest audit: May 28, 2026 at 1:57 AM UTC
• Version:
  1.0.3

Overview

This is a coherent project-management helper with local calculation and health-check commands, and I found no evidence of exfiltration, persistence, or hidden destructive behavior.

Install only if you want an AI assistant to help structure project-management workflows. Be aware it may inspect filenames and modification times in project folders you ask it to health-check, and it can read a risks JSON file you explicitly provide for scoring.

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: 89% confidence
• Finding: The skill description is unusually broad and user-invocable, covering essentially any project planning, tracking, risk, reporting, or management request. That can cause the skill to be selected in situations beyond the user's explicit intent, increasing the chance of inappropriate instruction injection into unrelated workflows or unexpected tool/config access.

Natural-Language Policy Violations

Low

• Confidence: 74% confidence
• Finding: Defaulting to a hybrid methodology when the user's intent is unclear causes the skill to make a substantive workflow decision without explicit confirmation. While not directly enabling code execution, it can steer planning outputs, file/template selection, and downstream recommendations in ways the user did not request, reducing reliability and potentially causing inappropriate project governance actions.

VirusTotal

66/66 vendors flagged this skill as clean.