Security audit

Proclaw Scam insight(骗局洞察者)

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly local anti-scam education, but it also teaches offensive market-control, social-engineering, insider-information, and investment-decision tactics that need review before installation.

Install only if you want a Chinese-language educational finance and scam-analysis pack and can review its advanced sections carefully. Treat outputs as educational, not financial advice, and do not use the social-engineering, pricing-power, insider-information, rule-design, leverage, or market-control material to manipulate others or make personalized investment decisions.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (18)

Context-Inappropriate Capability

Medium

Confidence: 88% confidence
Finding: This section shifts from defensive anti-scam guidance into teaching users how to become a market or structure designer with pricing and information advantages. In context, that can normalize manipulative or exploitative conduct and move the skill toward operational guidance for asymmetric influence rather than user protection.

Context-Inappropriate Capability

Medium

Confidence: 90% confidence
Finding: The advanced-capability section explicitly frames progression toward rule design, pricing power, and traffic control. That expands the skill from analysis into teaching strategic influence over markets or participants, which is misaligned with an anti-scam tool and could facilitate manipulative behavior.

Description-Behavior Mismatch

High

Confidence: 97% confidence
Finding: The document departs from the stated anti-scam purpose and explicitly reframes the user from a defender into a market designer who controls rules, pricing, and flow. In a scam-insight skill, this is dangerous because it normalizes offensive manipulation concepts and can operationally steer users toward exploitative market behavior rather than scam detection.

Context-Inappropriate Capability

High

Confidence: 98% confidence
Finding: This section provides a capability framework for manipulation—narrative design, traffic control, rule design, and price shaping—which can be directly repurposed to engineer pump-and-dump or fraud schemes. Because the skill is presented as anti-scam education, embedding offensive playbooks increases the risk of misuse under the cover of a safety-oriented tool.

Context-Inappropriate Capability

Critical

Confidence: 100% confidence
Finding: The text explicitly endorses using pseudo-logic, selective data, and cherry-picked cases to build 'interpretation power,' which is a textbook recipe for deceptive persuasion. In financial contexts, these tactics can facilitate fraud, market manipulation, and coerced investor decision-making, making this especially dangerous.

Description-Behavior Mismatch

High

Confidence: 96% confidence
Finding: The '获取定价权' section gives actionable guidance on acquiring information asymmetry, resource control, rule-setting power, and narrative dominance—capabilities often abused in financial manipulation. In context, this is not merely descriptive analysis; it reads as a roadmap for obtaining leverage over other market participants.

Context-Inappropriate Capability

High

Confidence: 99% confidence
Finding: Presenting a 'harvester model' with core logic centered on narrative design, risk transfer, traffic control, and extraction operationalizes a predatory scam structure. Even if framed analytically, the level of structured decomposition makes it easy to adapt as a how-to template for exploitation.

Description-Behavior Mismatch

High

Confidence: 97% confidence
Finding: The progression from participant to rule designer encourages the user to evolve toward influence, control, and market design rather than toward safer decision-making. In an anti-scam skill, this context amplifies risk because it can legitimize predatory ambition under the guise of learning scam mechanics.

Description-Behavior Mismatch

High

Confidence: 98% confidence
Finding: Defining 'offense' as mastering pricing power, designing rules, and controlling flow directly conflicts with the skill's claimed anti-harvest purpose and shifts the user toward manipulative conduct. This creates a clear misuse pathway by recasting exploitative control as a recommended next step.

Context-Inappropriate Capability

High

Confidence: 97% confidence
Finding: The document contains actionable, step-by-step scam enablement content such as trust-building phases, propagation strategies, information manipulation, and emotional control tactics framed as reusable algorithms. Although the broader skill is positioned as fraud analysis, this section provides sufficient operational detail for misuse, making it dangerous dual-use content that meaningfully lowers the barrier for conducting social engineering scams.

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The document explicitly reframes an anti-scam skill from defensive education into offensive market play, including upgrading from 'not being harvested' to 'actively designing' strategies. In the context of a scam-analysis skill, this broadens the skill into manipulation-oriented financial tactics and can normalize exploitative conduct beyond the declared purpose.

Context-Inappropriate Capability

High

Confidence: 99% confidence
Finding: These lines recommend obtaining 'industry insider news' and building networks with 'internal persons' to acquire information unavailable to others. In a financial context, this can encourage insider-style information gathering and misuse of non-public information, which is especially dangerous because the skill is presented as anti-scam rather than regulated compliance or legal education.

Context-Inappropriate Capability

High

Confidence: 98% confidence
Finding: The section teaches monetizing information advantages by positioning before public disclosure and profiting after release, as well as exploiting information asymmetries. That is incompatible with an anti-scam skill and can facilitate market abuse, unfair dealing, or behavior adjacent to insider trading and manipulative advisory practices.

Description-Behavior Mismatch

High

Confidence: 97% confidence
Finding: This portion instructs users to design rules, control flow, accumulate key resources, and gain pricing power, which moves from scam awareness into influence and control tactics. Within an anti-scam skill, such guidance is especially concerning because it can be repurposed to engineer asymmetric systems that exploit less-informed participants.

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The script’s behavior does not match the skill’s stated purpose of identifying financial scams and explaining fraud mechanisms; it only computes a Kelly criterion position size and then emits a binary investment recommendation. In a scam-analysis context, this mismatch is dangerous because users may treat the output as safety validation for an investment, creating false reassurance and potentially facilitating fraudulent schemes rather than detecting them.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The trigger conditions are broad enough to capture ordinary investment and financial-advice requests, not just scam detection. That increases the chance the skill will be invoked in high-stakes financial decision contexts where users may over-rely on its outputs without appropriate safeguards.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill provides concrete investment-analysis workflows, calculators, scoring systems, and decision outputs without a clear warning that results are not financial advice. In a financial context, quantitative outputs can strongly influence user decisions, so missing disclaimers and guardrails materially increase the risk of harmful reliance.

Missing User Warnings

High

Confidence: 93% confidence
Finding: The opening frames scam-related social engineering as a set of 'core algorithms' without an immediate warning that the material is hazardous, deceptive, and must not be used offensively. That framing normalizes the abusive techniques and increases the chance that readers interpret the document as an instructional guide rather than a defensive analysis resource.

Static analysis

No suspicious patterns detected.