Smart Prompt Generator

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:18 PM UTC
• Version:
  1.0.0

Overview

This is a prompt-template skill with no executable runtime behavior or hidden access, but users should review prompts before sharing them with outside AI tools.

Install only if you want a prompt-writing/template helper. Review generated prompts before use, do not include passwords, API keys, customer data, proprietary code, or other sensitive information, and treat the documented security controls as claims rather than proven technical enforcement.

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
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (8)

Intent-Code Divergence

Medium

• Confidence: 90% confidence
• Finding: This section makes concrete security and privacy claims such as sanitization, validation, and fully local processing, but the SECURITY.md file itself does not provide evidence that these controls actually exist or are consistently enforced. Overstated assurances can cause users and reviewers to trust generated prompts or the skill's handling of data more than warranted, increasing the chance of unsafe use or missed defects.

Intent-Code Divergence

Medium

• Confidence: 93% confidence
• Finding: The architecture diagram claims the template engine is 'sandboxed' and the generated output is 'safe to use,' which are strong guarantees not proven by this documentation alone. In a prompt-generation skill, such wording is risky because users may treat generated prompts as inherently trustworthy despite the possibility of prompt injection, unsafe template content, or incomplete filtering.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The bug template explicitly asks users to paste both their prompt and the generated output, but provides no warning to avoid including secrets, personal data, proprietary content, API keys, or sensitive model interactions. In a prompt-generation skill, users are especially likely to share full prompts and outputs verbatim, which can expose confidential information publicly through GitHub issues.

Vague Triggers

Medium

• Confidence: 91% confidence
• Finding: The activation guidance says to ask the skill for help creating a prompt, which is broad enough to match many ordinary user requests about prompts. In agent environments, overly generic activation criteria can cause unintended invocation, context leakage into unrelated tasks, or prompt-shadowing where this skill intercepts requests not meant for it.

Vague Triggers

Medium

• Confidence: 94% confidence
• Finding: The closing call-to-action, 'Just ask the skill to help create a prompt for your specific need,' encourages vague invocation without defining boundaries. This can increase accidental triggering and make the skill compete with other skills or system behaviors on common requests, especially in multi-skill agent routing.

Vague Triggers

Medium

• Confidence: 95% confidence
• Finding: The README instructs users to activate the skill with a very broad natural-language phrase such as asking for help creating a prompt. In agent systems that route skills based on semantic matching, this can cause unintended activation in unrelated contexts, exposing users to prompt injection surface or unexpected behavior when they did not intend to invoke this skill.

Vague Triggers

Medium

• Confidence: 91% confidence
• Finding: The skill description is broadly worded and can trigger on many generic requests such as help creating prompts or improving AI interactions, causing overbroad activation. While not directly malicious, this increases the chance the skill is invoked outside its intended scope, which can lead to prompt confusion, unintended instruction injection exposure, or inappropriate delegation of user requests.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The README explicitly instructs users to paste generated prompts into third-party AI tools but does not warn them to remove secrets, proprietary code, customer data, or other sensitive context first. In a prompt-generation skill, this omission is meaningful because the tool is specifically designed to help users package project details for external submission, increasing the likelihood of accidental data disclosure.

VirusTotal

64/64 vendors flagged this skill as clean.