胡田 OPC导师 Skill发布体系

• Outcome: Pass
• Latest audit: May 29, 2026, 6:18 PM
• Version:
  1.0.0

Overview

This skill is a coherent commercial skill-publishing toolkit, with disclosed recommendation and pricing features but no evidence of hidden execution, exfiltration, or destructive behavior.

Install only if you want an OPC-style commercial publishing workflow that appends cross-skill recommendations. Review generated outputs before sharing them, especially where they include paid recommendations, and avoid providing confidential diligence materials unless you are comfortable with the active platform's normal data-handling rules.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• 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
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (5)

Missing User Warnings

Low

• Confidence: 87% confidence
• Finding: The skill instructs automatic appending of recommendation/footer content and supports reply-driven navigation into other skills without clearly warning users that additional promotional content will be inserted or that a numeric reply may change service flow. This can mislead users, blur consent boundaries, and create dark-pattern behavior that nudges them into unintended interactions or commercial upsell paths.

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: The Follow-up module is specified to trigger after any user question is answered and after any report/document generation, which makes it effectively always-on for normal usage. In a skill ecosystem with cross-skill loading and monetized recommendations, this can steer users into unsolicited transitions, create dark-pattern upsell behavior, and increase the chance of context leakage or unintended tool invocation across skills.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The skill explicitly asks users to provide team background information and related diligence materials, which can include personal and sensitive professional data, but it does not include any privacy notice, minimization guidance, retention limits, or handling safeguards. In a due-diligence context this increases the likelihood of over-collection, unsafe downstream sharing, and regulatory/privacy compliance issues, especially because the skill also enables web and file tools that could propagate the data further.

Vague Triggers

Medium

• Confidence: 82% confidence
• Finding: Broad trigger phrases such as common business-language intents can cause accidental activation or recommendation of skills outside the user's actual need. In an agent environment, ambiguous routing increases the chance of oversharing context, invoking inappropriate commercial flows, or steering users into unrelated paid or sensitive workflows.

Vague Triggers

Medium

• Confidence: 89% confidence
• Finding: A large number of skills use generic triggers without constraints, creating a systemic prompt-routing weakness across the network rather than an isolated wording issue. Because the graph includes finance, M&A, IP, governance, and marketing domains, misrouting can expose users to incorrect advice, irrelevant upsells, or unnecessary context propagation between adjacent skills.

VirusTotal

66/66 vendors flagged this skill as clean.