Back to skill

Security audit

tech-evaluation

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Chinese-language technology/product evaluation assistant with no artifact-backed evidence of hidden execution, persistence, credential handling, or destructive behavior.

Install this if you want a structured Chinese workflow for technical or product selection. Be aware that its trigger list is broad, so it may activate for generic comparison or recommendation requests, and it may use web/GitHub research when current product information is needed.

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
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (8)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger set includes very generic terms like '推荐' and broad comparison intent phrases in the description, which can cause the skill to activate during ordinary conversation that is not actually requesting a tech-selection workflow. Over-broad activation is dangerous because it can hijack unrelated user requests, change response behavior unexpectedly, and increase exposure to prompt-injection-style content from arbitrary inputs.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The manifest description and output requirements are written as mandatory Chinese behavior, while the trigger list also includes English terms like 'evaluate' and 'benchmark'. This indicates the skill may enforce a specific language or locale without explicit user opt-in or a documented region-specific justification.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The manifest trigger entries are insufficiently scoped and contain ambiguous short phrases, making accidental invocation likely across many normal user messages. In an agent system, ambiguous routing is a real security and reliability issue because the wrong skill can seize control of the interaction and apply instructions, tools, or assumptions outside the intended context.

Content

No source excerpt is available for this finding.

Overly Broad Trigger

Low
Category
Trigger Abuse
Confidence
90% confidence
Finding

The trigger '选型' is too short and generic, so it may match many unrelated requests that merely mention selection or choosing. Short generic triggers increase the chance of unintended skill activation, causing incorrect instruction routing and expanding the attack surface for adversarial inputs.

Content

No source excerpt is available for this finding.

Overly Broad Trigger

Low
Category
Trigger Abuse
Confidence
90% confidence
Finding

The trigger '对比' is a common everyday term and can be present in many conversations unrelated to this skill's purpose. This makes unintentional activation plausible, which can misroute user requests and apply the skill's strong formatting and analysis instructions in the wrong context.

Content

No source excerpt is available for this finding.

Overly Broad Trigger

Low
Category
Trigger Abuse
Confidence
90% confidence
Finding

The trigger '比较' is highly generic and likely to appear in normal speech, making activation too permissive. While not directly enabling code execution, this broad routing behavior is unsafe in agent environments because it increases the chance of the wrong skill influencing a session.

Content

No source excerpt is available for this finding.

Overly Broad Trigger

Low
Category
Trigger Abuse
Confidence
93% confidence
Finding

The trigger '推荐' is especially broad because users request recommendations for nearly anything, not just technical products. This can cause frequent accidental invocation, making the skill much more likely to process unrelated prompts and potentially override more appropriate task handling.

Content

No source excerpt is available for this finding.

Overly Broad Trigger

Low
Category
Trigger Abuse
Confidence
94% confidence
Finding

The trigger 'vs' is extremely short, language-agnostic, and commonly used in many contexts, so it is prone to accidental matches. In a skill-routing system, this creates noisy activation and can redirect unrelated comparisons into a specialized workflow unexpectedly.

Content

No source excerpt is available for this finding.