Back to skill

Security audit

trip-coupon

Security checks for vulnerabilities and agentic risk

Overview

This skill fetches travel coupon links and QR codes from a remote API, with disclosed affiliate-style attribution but no evidence of hidden execution, credential theft, destructive behavior, or system persistence.

Before installing, be aware that using this skill will contact an external coupon API and may show third-party coupon links or QR codes intended for WeChat. The helper script stores an anonymous install identifier in `scripts/.install_id` and sends it for attribution, so privacy-sensitive users may prefer not to run the script or should delete/reset that file if they want to rotate the identifier.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • 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
Findings (9)

Rp1

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The README instructs users to run npx skills add moooai/trip-coupon without pinning a specific version of the tool being executed. Unpinned npx execution increases supply-chain risk because users may fetch and run whatever version is current at install time, including a compromised or malicious release of the CLI or a hijacked dependency path.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The trigger phrases are broad natural-language requests such as asking the assistant to help claim travel coupons, without clear boundaries on when the skill should activate or what it should avoid doing. This can cause accidental invocation or over-eager routing, which is especially relevant here because the skill leads users to external coupon links and tracks attribution parameters.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill repeatedly instructs users to open links or scan QR codes in WeChat, effectively steering users into a specific external platform without offering a neutral alternative or justification. In this context, that increases phishing, tracking, and coercive redirection risk because the skill returns unauthenticated third-party links and QR images that users are encouraged to trust and open in another app.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The trigger phrases include broad terms such as '机票优惠券', '门票优惠', and '旅行省钱', while the documented API only returns ride-hailing and hotel coupon bundles plus a generic 'more coupons' QR entry. This can cause the agent to invoke the skill for unrelated travel-shopping intents and route users toward third-party affiliate links or WeChat redemption flows that do not actually match the requested service.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The document instructs clients to call an unauthenticated external API and surface third-party coupon and QR-code URLs directly, but provides no warning that users will be sent to external services or that those services may collect tracking or personal data. In a skill context, this increases phishing, privacy, and trust-boundary risks because consumers may treat the links as platform-endorsed content without informed consent or validation guidance.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
81% confidence
Finding

Multiple guideline strings instruct users to open links in WeChat or scan with WeChat, and the parameter descriptions repeat that recommendation as if it were the expected mode of use. This imposes a platform-specific locale/ecosystem preference without documenting user choice or justification, which matches the natural-language policy concern for forced locale/platform behavior.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The script creates a persistent install identifier, stores it locally, and then sends it to a remote API on every request for attribution. Even if the stated purpose is channel attribution, this still enables long-term request correlation and user/device tracking without clear disclosure, minimization, or consent.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The code automatically attaches a persistent installation ID as query parameters to all API requests with no user-facing notice or confirmation. This creates a privacy/security risk because the remote service can link repeated activity to the same installation over time, which is especially concerning in a skill context where users may not expect background identifiers to be transmitted.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The script's natural-language documentation, CLI description, and output strings are entirely in Chinese, and there is no indication that the skill is intentionally limited to Chinese-speaking users or a China-specific deployment. This can violate language/locale policy when a skill forces a specific language without user opt-in.

Content

No source excerpt is available for this finding.