Security audit

Fanli

Security checks for vulnerabilities and agentic risk

Overview

This shopping rebate skill is purpose-aligned and discloses its API/key requirements, but users should understand that product links are sent to a third-party service and converted links may be promotional links.

Install only if you trust feima-lab/fenxianglife and the separate fx-base dependency. Use a revocable API key, avoid submitting shopping links that contain private account/session tokens, and treat generated purchase URLs as promotional or affiliate-style links.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Context-Inappropriate Capability

Medium

Confidence: 90% confidence
Finding: This runner provides generic local capability discovery (`list`, `show`) and arbitrary sub-script execution (`call <接口名>`) for any `.mjs` file in the scripts directory, which goes beyond the narrowly described shopping/price-comparison purpose. In an agent-skill context, broad dispatch surfaces can expose unintended functionality, increase the blast radius of any malicious or risky script added later, and make it easier for users or upstream agents to invoke capabilities the skill did not explicitly promise.

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The README says users can paste product, Tao password, or Meituan links directly and the skill should activate without explicit commands, which makes invocation criteria very broad. In practice this can cause the agent to trigger on ordinary shopping links shared for discussion and automatically send link and shopping context to external price-comparison or affiliate services without clear user intent.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The README describes converting links, checking coupons, comparing prices, and querying historical prices, but it does not warn that user-submitted links and related shopping data are likely sent to Feima or other external services. This lack of transparency creates a privacy and consent risk, especially because affiliate, coupon, and cross-platform comparison workflows often involve transmitting URLs, identifiers, and possibly user shopping intent to third-party APIs.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The skill advertises activation on many broad, everyday shopping phrases such as '优惠', '便宜', '划算', and '美团', which can cause the agent to invoke the skill in loosely related conversations rather than only when the user clearly wants link conversion or price lookup. Because the skill sends user-provided shopping links to an external API, over-triggering increases the chance of unnecessary third-party data disclosure, affiliate-link rewriting, and unintended actions in contexts the user did not explicitly request.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The trigger example "帮我看看值不值得买" is overly broad and can match ordinary conversational requests without requiring a product link or clear shopping context. In an agent-routing system, this can cause accidental invocation of the shopping/conversion skill on unrelated user inputs, leading to incorrect tool use, unnecessary external lookups, or unintended handling of user content.

Static analysis

Detected: suspicious.dangerous_exec

Shell command execution detected (child_process).

Critical

Code: suspicious.dangerous_exec
Location: scripts/run.mjs:32