Promptify Skill

• Outcome: Pass
• Latest audit: May 27, 2026, 9:37 PM
• Version:
  3.1.0

Overview

This is a prompt-optimization skill with disclosed optional codebase reading, web research, and clipboard-helper output, with no evidence of hidden persistence, credential handling, destructive actions, or exfiltration.

Safe to install as a prompt helper if you are comfortable with it inspecting relevant project files when +deep or auto-detection selects codebase research, and with it using web search when +web or latest/best-practices language is present. Review any generated prompt before sharing it, and inspect the pbcopy command before running it so you do not overwrite the clipboard or execute badly quoted shell text.

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 (6)

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: The auto-detection example uses a very generic phrase ('Help me with this thing') to trigger clarifying-question behavior, showing that routing is based on broad natural-language heuristics. In a prompt-optimization skill that can invoke sub-agents, broad triggers can cause unintended capability activation, increasing the chance of unnecessary data collection or workflow changes from ordinary user text.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The README advertises web search, image analysis, conversation-context use, and clipboard copying without clear notice about what data may be accessed, transmitted, or stored. In a prompt-processing skill, these privacy-impacting features can expose sensitive prompts, repository content, screenshots, or prior chat context to external systems or local clipboard history without informed user consent.

Vague Triggers

Medium

• Confidence: 89% confidence
• Finding: The skill description uses broad activation phrases like 'improve this prompt' and 'make this clearer,' which can overlap with many ordinary user requests and cause the skill to trigger unintentionally. Because this skill can then auto-route to clarifying, codebase, or web-research sub-agents, accidental invocation may expand the task scope, expose extra context, or cause unnecessary tool use beyond what the user explicitly intended.

Vague Triggers

Medium

• Confidence: 92% confidence
• Finding: The auto-detection signals are vague terms such as 'best practices,' 'make it better,' or references to a project, which are common across many unrelated conversations. This ambiguity can misroute the skill into codebase or web-research workflows without clear user consent, increasing the chance of over-broad data access, unnecessary external lookups, or incorrect task delegation.

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The trigger conditions are broad enough to match many ordinary coding requests that mention an existing project, API, or modification task. This can cause the agent to invoke codebase exploration more often than intended, increasing unnecessary access to repository context and creating a larger chance of exposing sensitive code or instructions to downstream prompt-processing flows.

Missing User Warnings

Low

• Confidence: 94% confidence
• Finding: The skill directs copying generated content to the system clipboard (`echo 'PROMPT' | pbcopy`) without explicit user consent or warning. Automatic clipboard writes can overwrite sensitive clipboard contents and may cause users to paste unintended data into other applications, especially because the generated prompt may contain user-provided or context-derived sensitive information.

VirusTotal

66/66 vendors flagged this skill as clean.