Back to skill

Security audit

Token消耗优化器

Security checks for vulnerabilities and agentic risk

Overview

This is a markdown-only token-cost optimization skill with some broad activation and language-preference notes, but no hidden code, privileged actions, persistence, or exfiltration behavior.

Installers should expect this skill to give token-cost optimization advice, mostly in Chinese, and should verify any dated model-price table before relying on budget estimates. Its broad triggers may activate during adjacent cost or model-selection discussions, but the reviewed artifacts do not show hidden execution, data transfer, or persistent changes.

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 (5)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The trigger list is very broad and includes common cost-related phrases such as '成本优化' and '模型切换', which can cause the skill to activate in conversations where the user did not intend to invoke a token-auditing workflow. Unintended activation can inject irrelevant optimization guidance, override more appropriate skills, or expose internal budgeting/routing advice in the wrong context.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The activation conditions describe many general situations like high API cost, long prompts, model comparison, and budget planning, but they do not clearly define when the skill should not run or how to distinguish it from adjacent skills. This increases the chance of accidental invocation, leading to misrouting, unnecessary context loading, and potentially degraded system behavior in multi-skill environments.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The markdown explicitly states “中文输出,” which forces a specific language for responses. This is a natural-language policy issue because it does not provide user opt-in or explain a necessary region-specific reason for the restriction.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

Line L070 directs '中文创作/翻译' tasks to specific Chinese-optimized models, which is a language-based routing policy stated as a default behavior. Because the document does not offer user opt-in or explain a region-specific justification, this can violate the language/locale policy requirement.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

Lines L083-L086 use is_chinese(task_type) to choose a different model path for Chinese content, embedding a language-based policy directly into the example logic. The file provides no indication that the user can choose this behavior or that the locale constraint is otherwise justified.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.