Back to skill

Security audit

guide-generator

Security checks for vulnerabilities and agentic risk

Overview

This guide-writing skill is mostly coherent, but it tells the agent to use local PowerShell curl for web retrieval without clear safeguards.

Install only if you are comfortable with a guide skill that may ask to save Markdown files locally. Avoid letting it use PowerShell or curl for article retrieval unless the exact URL and command are visible and you approve them.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The file frames the skill as a Chinese攻略 generator, mandates Chinese content structure and wording conventions, and primarily defines operation in Chinese, while also listing some English triggers. It does not explicitly offer users a choice of output language or explain that the Chinese-language constraint is optional or region-specific.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger list includes broad phrases such as "规划 XX 行程", "周末去哪儿", "找好吃的", "XX 值得去吗", and English phrases like "food recommendations in [city]" that are common conversational requests rather than narrowly scoped invocation phrases. The file does not provide exclusion conditions or negative examples to clarify when the skill should not activate.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill explicitly instructs use of local PowerShell via curl.exe to fetch WeChat articles, which expands behavior from web research into host-level command execution. Even if the stated goal is content retrieval, invoking OS commands creates a dangerous bridge where attacker-influenced URLs or future prompt injections could lead to unintended local command use, data access, or abuse of system capabilities.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill defines a default behavior to save output to a local filesystem path, despite the manifest primarily describing guide generation rather than local file operations. Default writes to a fixed path can cause unintended data persistence, privacy issues, overwriting of user files, or misuse if prompt-influenced filenames or destinations are later introduced.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill omits upfront disclosure that generated content may be written to local storage by default. This undermines informed user consent and increases the chance of silent persistence of potentially sensitive travel plans, preferences, or other personal content on the host machine.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The template library is entirely written to produce Chinese-language outputs and does not indicate any user-language detection, preference check, or opt-in. This can override or ignore a user's preferred language, reducing usability and potentially causing misunderstanding of safety, travel, or logistical details, though it is not an exploit in the classic code-execution sense.

Content

No source excerpt is available for this finding.