Back to skill

Security audit

Flight Price Advisor

Security checks for vulnerabilities and agentic risk

Overview

This is a markdown-only flight price advice skill with disclosed SerpAPI setup; the main concerns are broad trigger wording and minor quality issues, not malicious behavior.

Use this if you want Chinese-language flight price advice. Configure SerpAPI only in the intended project, prefer a limited API key, and be aware the skill may activate on generic price questions unless its triggers are narrowed.

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
  • 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 (6)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The manifest patterns contain ambiguous regex-like expressions for common questions about price, value, and whether to buy now, without anchoring them to flights. Because the skill has high priority and no context isolation, these patterns increase the chance of intercepting unrelated user requests, causing scope confusion and potentially exposing users to irrelevant external-key setup instructions or unintended workflow activation.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The trigger scenarios are broad enough to match generic price and purchase-timing questions such as '价格怎么样' or '现在买还是再等等', which are not inherently flight-specific. In a shared agent environment, this can cause unintended invocation on unrelated shopping or pricing conversations, leading to wrong-skill responses, unnecessary guidance to configure external services, or accidental access attempts to flight-price data paths.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The file title and description define the response templates entirely in Chinese and present them as the templates the AI should use when generating responses. There is no indication that users may choose another language or that the Chinese-only constraint is optional, which constitutes a language/locale policy concern.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

This markdown file presents all instructions, labels, warnings, and examples exclusively in Chinese, and nowhere indicates that the skill is China-specific or that users may choose another language. The policy explicitly flags language or locale constraints when they are forced without user opt-in or justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

The README presents the skill purpose and usage entirely in Chinese, which effectively forces a specific language for users without any opt-in or alternative. Under the policy, language constraints should be optional or clearly justified as region-specific.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The document defines 'low' as more than 15% below the average price (L051-L054), and the sample calculation shows pctDiff as -10.4 at L214. Despite that, the example labels the price level as 'low' at L215, which directly contradicts the stated algorithm rather than merely omitting detail.

Content

No source excerpt is available for this finding.