Back to skill

Security audit

多平台内容检测&安全改写

Security checks for vulnerabilities and agentic risk

Overview

This skill is a Chinese content compliance and rewriting helper with disclosed web-search and reference-reading behavior, but users should notice its local-file wording is inconsistent.

Install only if you want China-focused content compliance assistance. Treat its output as drafting support, not legal approval, and do not confirm reading a local file unless you intentionally want that exact file used as the text to check.

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
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (9)

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The manifest explicitly states the skill does not read user local files, but the workflow later permits reading a user-specified file path after confirmation. This mismatch is security-relevant because it weakens user trust boundaries and can lead to unintended disclosure of sensitive local data if the agent or user assumes file access is impossible.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

The description states the skill operates on ‘任意中文文本’ and the workflow is built around Chinese-platform rules, while also prescribing behavior for English text only in a limited fallback mode. This imposes a language/locale constraint without clearly offering the user a language choice or documenting a region-specific compliance justification in policy terms.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The trigger conditions are broad enough to activate on many generic editing or review requests, which increases the chance the skill runs when the user did not intend this specific workflow. Because the skill mandates web searches and extensive policy checks, unintended invocation can cause unnecessary data exposure to external services, altered outputs, or surprising behavior.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The workflow text authorizes local file reads despite the top-level description promising that no local files are read. An attacker or careless operator could exploit this inconsistency to socially engineer a user into approving reads of sensitive files under a false expectation that the skill never accesses local data.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The description states the skill operates on “任意中文文本,” which is a natural-language constraint to a specific language. Because the manifest does not present this as an explicit user-selectable option or justified region-specific limitation, it may violate language/locale policy expectations.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The trigger phrases are generic everyday terms such as '内容检查' and '广告法检查', which can activate the skill during broad, unrelated user requests. Because this skill claims to perform mandatory live web searches and content rewriting, overbroad invocation increases the chance of unintended routing, unnecessary external access, and silent modification of user content in contexts where the user did not explicitly request this specific skill.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

Line L003 states the checklist is adapted specifically for Chinese writing scenarios, and the document throughout prescribes mandatory stylistic rules in that language context. Because the file presents these constraints as universal delivery requirements rather than an opt-in or region-specific policy, it can be read as forcing a specific language/locale standard.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

This file imposes a single language for all guidance and output instructions, including headings and procedural text, but does not indicate that Chinese is optional or required for a specific region-only workflow. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The title and introductory description present the skill content entirely in Chinese and frame it as the default operating language for platform-rule checking. Under the policy, forcing a specific language without user opt-in can be a natural-language locale violation unless the restriction is explicitly justified as region-specific.

Content

No source excerpt is available for this finding.