ProClaw Top Design Prompts(顶级平面设计提示词专家)

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:42 PM UTC
• Version:
  1.0.1

Overview

This is a markdown-only design prompt skill with no device access, but users should review its persuasion and social-pressure guidance before using it for public campaigns.

Safe to install from a device-security perspective: it is instruction-only and asks for no credentials or system access. Before using outputs publicly, review generated prompts for fabricated popularity, false urgency, exaggerated authority or safety claims, and pressure tactics, especially in health, finance, education, mental health, minors, or other vulnerable-audience contexts.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• 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

Findings (23)

Context-Inappropriate Capability

Medium

• Confidence: 96% confidence
• Finding: The document explicitly frames the system as using '情感驱动' and '自动触发情感反应并驱动行动', which goes beyond neutral prompt generation into persuasive influence design. In the context of a broadly applicable design agent serving 50+ industries, this can be used to generate manipulative promotional materials that steer audience behavior without appropriate safeguards or disclosure.

Context-Inappropriate Capability

High

• Confidence: 99% confidence
• Finding: This section explicitly includes social-pressure tactics such as '朋友圈都在参加，你还在等什么' as an action trigger, which is a classic coercive persuasion mechanism unrelated to legitimate design assistance. Embedding such tactics in an automated generation pipeline enables scalable creation of manipulative content that exploits conformity pressure to induce action.

Context-Inappropriate Capability

High

• Confidence: 98% confidence
• Finding: The optimization loop is designed to repeatedly strengthen emotional and action intensity until thresholds are met, turning the skill into an automated persuasion optimizer rather than a neutral design helper. This materially increases misuse potential because it systematizes iterative refinement of manipulative influence at scale.

Context-Inappropriate Capability

Medium

• Confidence: 95% confidence
• Finding: This section explicitly instructs the agent to use urgency, scarcity, value framing, and social pressure to drive action, which are manipulative persuasion tactics rather than neutral design assistance. In the context of a prompt-engineering skill, operationalizing pressure-based influence can enable coercive marketing or behavioral manipulation at scale and without user awareness.

Context-Inappropriate Capability

Medium

• Confidence: 84% confidence
• Finding: The document introduces biometric and neuro-measurement methods such as eye tracking, EEG, and fMRI that are unnecessary for a design-prompt skill and expand it toward sensitive psychological optimization. Even if not directly executable, this normalizes collecting or optimizing against highly sensitive signals without clear necessity, consent, or governance.

Description-Behavior Mismatch

Medium

• Confidence: 96% confidence
• Finding: The file states that its mission is to trigger emotions and drive users to act, which materially exceeds the declared purpose of generating high-quality design prompts. That goal reframes the skill as a behavioral influence system, increasing the risk that outputs are optimized for manipulation rather than user-serving creative assistance.

Context-Inappropriate Capability

Medium

• Confidence: 96% confidence
• Finding: The skill goes beyond generating neutral design prompts and explicitly aims to shape user behavior through emotional triggering and action-driving mechanisms. In this context, embedding persuasion tactics such as trust-building, desire stimulation, and signup pressure creates a manipulative marketing engine rather than a purely creative assistant, increasing the risk of covert influence on end users.

Context-Inappropriate Capability

Medium

• Confidence: 94% confidence
• Finding: The skill infers demographic, income, education, psychology, motivations, fears, and emotional vulnerabilities from minimal user input without justification or verification. This creates a profiling capability that can be inaccurate, biased, and then used to tailor manipulative messaging toward presumed weaknesses or sensitivities.

Context-Inappropriate Capability

High

• Confidence: 99% confidence
• Finding: This section explicitly prescribes social-pressure persuasion such as implying that peers are already participating to push action. Social proof can be legitimate when factual, but here it is framed as a generic behavioral trigger, which enables deceptive or coercive messaging unrelated to the legitimate purpose of producing a design prompt.

Context-Inappropriate Capability

Medium

• Confidence: 95% confidence
• Finding: The emotion-trigger layer is operationalized with target intensity thresholds and optimization loops for emotional arousal and action conversion. That transforms the skill into a system for systematically increasing susceptibility and response rates, which is materially more dangerous than ordinary prompt generation because it encourages iterative refinement of manipulative influence.

Vague Triggers

Medium

• Confidence: 91% confidence
• Finding: The trigger conditions are framed so broadly that the skill could activate for a wide range of ordinary design-related requests, increasing the chance of unintended invocation and overreach. In an agent ecosystem, overly broad activation can cause the wrong skill to take control, produce manipulative brand-output workflows, and bypass more appropriate narrower skills or user intent checks.

Vague Triggers

Medium

• Confidence: 94% confidence
• Finding: The manifest description claims near-universal industry coverage and minimal-input automatic completion with no meaningful limiting constraints, which makes the skill prone to over-selection and unchecked inference about user goals. This is dangerous because the agent may autonomously fill in strategic, emotional, and persuasive content the user did not explicitly request, creating both scope creep and potential manipulation.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The markdown describes emotional manipulation and social-pressure techniques without any warning, consent mechanism, or limitations on use, despite clearly intending to influence behavior. Absent safeguards, users may deploy generated content that is coercive, deceptive, or inappropriate for sensitive contexts.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The document promotes emotional triggering and action-driving without any disclosure that these are potentially manipulative techniques that can impair user autonomy. Lack of warnings or ethical boundaries makes it more likely that operators will deploy covert pressure tactics as normal design practice.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: This section specifically endorses urgency and social-pressure techniques such as limited seats, countdowns, and 'everyone is joining' messaging, but gives no warning about their coercive effects. In a scalable AI skill, these tactics can be mass-produced and tailored, amplifying manipulative influence over decision-making.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: The overview explicitly highlights emotional triggering and action-driving as core value without any user warning, consent mechanism, or ethical boundary. Presenting covert persuasion as a standard capability is dangerous because users may deploy the skill for manipulative campaigns while recipients are not aware that the content was optimized to exploit trust, desire, and urgency.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The architecture includes an 'action trigger layer' with social-pressure mechanisms as a core system component, again without any caution or ethical guardrails. This normalizes pressure-based influence as part of the design pipeline and makes misuse easier because the manipulative behavior is built into the workflow rather than added incidentally.

Ssd 4

Medium

• Confidence: 97% confidence
• Finding: This workflow explicitly stages trust, craving, reassurance, and social pressure to drive action, constituting a structured persuasion pipeline rather than ordinary creative assistance. In a general-purpose design skill, such staged influence mechanisms increase the risk of manipulative marketing content, especially when applied across industries and audience types without safeguards.

Ssd 4

Medium

• Confidence: 98% confidence
• Finding: The system measures emotional and action intensity, then iterates until influence thresholds are exceeded, which is an explicit optimization loop for behavioral manipulation. This is dangerous because it provides a reusable recipe for systematically increasing persuasive pressure beyond what users may recognize as ethical marketing.

Ssd 4

Medium

• Confidence: 94% confidence
• Finding: The staged sequence explicitly describes a funnel that first attracts attention, then builds desire, lowers concerns, establishes trust, and finally drives action. This is dangerous because it systematizes resistance-lowering psychological influence in a way that can be reused across domains, including high-stakes or vulnerable-user contexts.

Ssd 4

Medium

• Confidence: 96% confidence
• Finding: The worked example operationalizes the manipulation framework by combining authority cues, safety framing, scarcity, value messaging, and social pressure into a concrete conversion playbook. Because it is example-driven and easy to reuse, it materially increases the likelihood of harmful deployment in real campaigns.

Ssd 4

Medium

• Confidence: 96% confidence
• Finding: This section lays out a staged persuasion pattern that deliberately combines authority cues, scarcity, safety, trust, and social pressure to drive conversions. In context, the cumulative structure is more dangerous than any single phrase because it provides a repeatable playbook for psychologically steering recipients toward signups.

Ssd 4

Medium

• Confidence: 95% confidence
• Finding: The workflow operationalizes progressive emotional conditioning, then validates and iterates until emotional and action thresholds are met. In this skill context, that makes the system more dangerous because it is not merely describing persuasive language, but instituting an optimization process for behavioral influence and signup pressure.

VirusTotal

64/64 vendors flagged this skill as clean.