ClawHub
Back to skill

Security audit

Renxing Shangzhan

Security checks for vulnerabilities and agentic risk

Overview

The skill does not show malware-like system access, but it broadly auto-triggers and provides manipulative marketing guidance based on audience fears and weaknesses, including vulnerable groups.

Review this carefully before installing. It appears locally contained, but it is designed to generate persuasion and conversion advice from psychological weaknesses. Use it only for ethical analysis or defensive review, avoid using it for targeting minors or vulnerable people, and narrow the triggers before publishing or enabling it broadly.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (10)

Vague Triggers

High
Confidence
97% confidence
Finding
The trigger list contains very broad terms such as '分析', '营销', and 'analyze' that are likely to activate on many unrelated user requests. This can cause the skill to hijack normal conversations and inject guidance about psychological triggers and business manipulation into contexts where the user did not explicitly request it.

Missing User Warnings

High
Confidence
99% confidence
Finding
This section actively generates targeted manipulation guidance based on audience fears, desires, and weaknesses, including minors and other potentially vulnerable groups, and recommends coercive conversion tactics such as exploiting loss aversion, anxiety, vanity, sunk cost, and social pressure. In the context of a user-facing agent skill, this meaningfully facilitates psychological exploitation, deceptive marketing, and abusive persuasion at scale without warnings, constraints, or ethics guardrails.

Natural-Language Policy Violations

Medium
Confidence
93% confidence
Finding
The skill explicitly frames itself as a 'business warfare manual' and provides analysis of greed, vanity, sunk cost, and other human weaknesses without any safety boundaries or ethical constraints. In context, this enables optimization of manipulative persuasion tactics against users or customers, increasing the risk of fraud, coercive marketing, or exploitative targeting.

Natural-Language Policy Violations

Medium
Confidence
77% confidence
Finding
The manifest value `renxing-shangzhan` is a natural-language identifier in Chinese pinyin, which suggests the skill may be tied to a specific language or locale. In this file there is no accompanying description, opt-in, or justification indicating that the locale restriction is intentional or user-selectable.

Overly Broad Trigger

Low
Category
Trigger Abuse
Confidence
91% confidence
Finding
The trigger '人性' is extremely short and generic, so the skill may activate for broad philosophical, literary, psychological, or social discussions unrelated to this skill. Because the skill's content centers on exploiting human weaknesses, accidental invocation increases exposure to manipulative guidance in benign contexts.

Overly Broad Trigger

Low
Category
Trigger Abuse
Confidence
90% confidence
Finding
The trigger '诱因' is too broad and can match many harmless discussions about causes, incentives, or motivations. This raises the chance of unintended activation of a skill that discusses leveraging psychological triggers, which is risky given the manipulative framing of the content.

Overly Broad Trigger

Low
Category
Trigger Abuse
Confidence
98% confidence
Finding
The trigger '分析' is among the broadest possible activators and may match a large share of ordinary requests. This makes the skill especially prone to unsolicited activation, and because the skill promotes analysis of human weaknesses for commercial strategy, the context makes the overbreadth more dangerous than a typical generic analysis skill.

Overly Broad Trigger

Low
Category
Trigger Abuse
Confidence
92% confidence
Finding
The trigger '营销' is generic and likely to overlap with many legitimate marketing questions. In this skill, broad activation is risky because responses may steer users toward exploiting psychological weaknesses rather than ethical marketing practices.

Intent-Code Divergence

Low
Confidence
96% confidence
Finding
The help section says users should invoke commands as `analyze`, `fusion`, and `advisor`, but the code actually dispatches only on `--analyze`, `--fusion`, `--advisor`, or `-a` at L126, L162, and L203. This is an active documentation-versus-code contradiction rather than a mere omission, because the documented forms will not trigger the advertised functionality.

Natural-Language Policy Violations

Low
Confidence
95% confidence
Finding
This code presents all usage/help text in Chinese only, and the rest of the script's output follows the same fixed locale pattern. Under the policy for natural-language violations, forcing a specific language without user opt-in is a reportable issue unless the locale restriction is documented and justified.

Static analysis

No suspicious patterns detected.