Security audit

AHTV PK to Xunlei

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AHTV episode resolver and Xunlei cloud-save workflow, with expected cloud-login risk but no hidden, destructive, or unrelated behavior found.

Install only if you want an agent to access AHTV and interact with your Xunlei Cloud Drive for this specific show. Treat your phone number and SMS code as sensitive, provide them only in the real Xunlei login flow, and review any cloud-drive add, move, or rename action before it is completed.

A.I.G

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

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (7)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 98% confidence
Finding: The code substantially matches the discovery/scraping portion of the description: it parses single dates, lists, closed ranges, and open ranges; finds 快乐无敌大PK episode candidates on ahtv.cn; narrows candidates; extracts a real media URL; and formats output filenames and target cloud paths. However, the declared purpose says the skill saves episodes into 迅雷云盘, dedupes existing 迅雷云盘 files, and uses a specific SMS login method. None of those behaviors appear in the supplied code. The code only resolves metadata and URLs and returns structured results. Therefore the description overstates key capabilities and is materially inaccurate.
Content: No source excerpt is available for this finding.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 97% confidence
Finding: The declared description presents a full media-retrieval-and-cloud-save skill centered on 快乐无敌大PK episodes and 迅雷云盘 integration. In contrast, this code chunk only performs date-expression parsing and latest-date resolution, then prints structured output. While date parsing is a supporting piece of the declared workflow, the actual chunk does not implement the primary advertised behavior or the key external-resource interactions (video discovery, URL extraction, cloud storage access, deduplication, login). Therefore the description does not accurately represent what this supplied code chunk actually does.
Content: No source excerpt is available for this finding.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 97% confidence
Finding: The declared description promises a complete workflow: discover episode pages from date expressions, extract real video URLs, then save episodes into 迅雷云盘 with deduplication and specific naming, using a specified SMS-login method. The provided code chunk implements only the discovery/resolution portion. It parses CLI arguments, computes the applicable dates, calls library helpers to resolve dates/episodes through AhtvClient, and emits structured JSON. There is no code for cloud drive access, authentication, file existence checks, upload/download, or output naming. While resolving episodes may be a supporting part of the declared workflow, the major advertised capabilities around 迅雷云盘 are absent from this code, so the description materially overstates what this code chunk actually does.
Content: No source excerpt is available for this finding.

Undeclared Tool Scope

Medium

Category: MCP Least Privilege
Confidence: 91% confidence
Finding: The skill instructs the agent to access external websites, search AHTV, open Xunlei Cloud Drive, and perform cloud-side link ingestion, but it declares no explicit tool scope or permissions boundary. Missing scope metadata increases the chance of overbroad network/browser use and makes review, sandboxing, and policy enforcement weaker.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The skill directs SMS-based login to Xunlei Cloud Drive using a phone number and one-time SMS code, but provides no warning, consent language, or handling constraints for these sensitive authentication factors. This can lead to unsafe collection, exposure, or misuse of credentials and increases phishing-like risk in an automation context.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 92% confidence
Finding: The default request headers hard-code Accept-Language to prefer zh-CN, which is a natural-language locale policy choice embedded in code. The file does not offer user opt-in or configurability for this locale preference, and no region-specific justification is documented in the file.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The argument help string contains Chinese text ("快乐无敌大PK") in an otherwise English-language CLI without offering any language or locale choice. This is a natural-language policy concern because it imposes a locale-specific user-facing string without opt-in or justification.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.