Back to skill

Security audit

1lou Auto Follow

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Chinese-language TV update checker/downloader, but it is tightly tied to one user's local paths and qBittorrent setup.

Install only if you want an agent to read the disclosed watchlist file, contact the listed 1lou domains, and, after confirmation, help add torrent files to the listed qBittorrent service. Review the hard-coded macOS path and private IP addresses first because they appear tailored to a specific user's machine.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (8)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding

The skill clearly instructs use of shell-capable actions such as running curl and performing local file operations, but it declares no explicit tool scope or permissions boundary. In an agent environment, undocumented shell access increases the chance of over-privileged execution, unsafe command use, and user-unexpected side effects such as downloading files or modifying local state.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger condition includes broad user queries related to '追剧' functionality, which can cause the skill to activate in situations where the user did not intend network access, local file reads, or downstream download behavior. Because this skill also touches watchlists, local files, and qBittorrent, overbroad invocation materially raises the risk of unintended side effects.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The stated purpose is checking and pushing episode updates, but the skill also performs torrent acquisition, local filesystem searches, copying files, and qBittorrent control. This expands the trust boundary significantly: a user invoking a benign-seeming update checker could trigger downloads, persistence changes, and interaction with local infrastructure services.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The manifest says the skill will '自动检查并推送1lou网站的最新剧集更新', which implies both checking and delivering notifications. The implementation does check the site and build a markdown message, but it only returns the string and logs it via console, with no code to call any notification, messaging, or delivery mechanism.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill emits all visible status and result strings in Chinese, including console output and the final message, without offering any language selection. That creates a natural-language policy concern if skills are expected not to force a locale without user opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

The entire skill description and prescribed user-facing responses are written only in Chinese, with no indication that the user can choose another language or that the skill is intentionally limited to Chinese-speaking use. This creates a locale/language policy issue because the skill implicitly enforces one language without opt-in.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
77% confidence
Finding

The surrounding code and comments suggest this parser is intended to work for arbitrary watchlist keywords, especially since main() iterates over a watchlist and passes item.keyword into parse(). However, one of the declared regex patterns is hard-coded to '不良医生', which conflicts with the documented/generalized parsing intent and indicates the implementation is partly specialized despite generic comments.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
77% confidence
Finding

This code accesses a user-specific file at a fixed home-directory path to read watchlist data, but there is no comment, docstring, or explicit user-facing warning that local user data will be read. The existing logs only announce the check process and domain selection, not that the skill inspects a local file in the user's home directory.

Content

No source excerpt is available for this finding.