Software Manager Skill

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:42 PM UTC
• Version:
  2.2.0

Overview

This is a coherent product-management helper with disclosed web research, local reference use, and optional export scripts; I found no hidden credential use, persistence, purchases, or destructive behavior.

Install only if you are comfortable with product details being used in web searches, optional Node/Python package installs for export features, and user-directed file writes. Choose export paths carefully and review generated documents or H5 prototypes before sharing; the H5 instructions contain mixed claims about zero external dependencies versus Tailwind CDN use.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Output Handling: Unvalidated Output Injection, Cross-Context Output, Unbounded Output
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: The trigger list is broad enough that the skill may activate for generic product, agile, or documentation queries without clear user intent to invoke this capability. In an agent environment, over-broad activation can cause unintended use of web search and local document access, expanding data exposure and causing the model to take actions the user did not explicitly request.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The README advertises multi-source search and local knowledge-base access but does not clearly warn users that the skill may read local files and query the network as part of its workflow. This undermines informed consent and can lead to privacy or data-handling surprises, especially if prompts or generated content include sensitive business information.

Missing User Warnings

Low

• Confidence: 86% confidence
• Finding: The workflow explicitly performs network searches, but the description does not warn users that external queries may be made. This matters because user prompts may contain sensitive project details, and sending them to search or fetch tools without clear notice can leak confidential information.

Unvalidated Output Injection

High

• Category: Output Handling

Content

os.unlink(mmd_path)
            return False

        result = subprocess.run(
            ['node', mm_script, mmd_path, output_path],
            capture_output=True, timeout=60
        )

• Confidence: 80% confidence
• Finding: subprocess.run( ['node', mm_script, mmd_path, output_path], capture_output

VirusTotal

65/65 vendors flagged this skill as clean.