Back to skill

Security audit

AI 文本净化器(中文版)

Security checks for vulnerabilities and agentic risk

Overview

The skill is a local Chinese text detector and rewriter, but it is explicitly designed to reduce AI-detection signals and remove AI-origin traces.

Install only if you are comfortable with a tool whose core purpose includes making AI-generated Chinese text appear more human and reducing detectable AI-origin markers. Use it for transparent editing and proofreading, not to misrepresent authorship, bypass academic rules, or conceal provenance.

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
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (17)

Tp4

High
Category
MCP Tool Poisoning
Confidence
91% confidence
Finding

This variant includes undeclared CLI/file I/O behavior, which is security-relevant because agents may invoke the skill under incorrect assumptions about side effects. Hidden or insufficiently declared read/write behavior can expose local files or create/overwrite outputs unexpectedly, especially when paired with broad agent file access.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

This variant includes undeclared CLI/file I/O behavior, which is security-relevant because agents may invoke the skill under incorrect assumptions about side effects. Hidden or insufficiently declared read/write behavior can expose local files or create/overwrite outputs unexpectedly, especially when paired with broad agent file access.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

This variant includes undeclared CLI/file I/O behavior, which is security-relevant because agents may invoke the skill under incorrect assumptions about side effects. Hidden or insufficiently declared read/write behavior can expose local files or create/overwrite outputs unexpectedly, especially when paired with broad agent file access.

Content

No source excerpt is available for this finding.

Direct Prompt Extraction

High
Category
System Prompt Leakage
Confidence
85% confidence
Finding

Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.

Content

Scanner excerpt · core/rules.py (reported line 123)May include surrounding context.

python
raise FileNotFoundError(f"Rules file not found: {path}")
    with open(path, "r", encoding="utf-8") as f:
        data = json.load(f)
    return RuleSet(data)


def merge_rules(base: RuleSet, user_rules: Dict[str, Any]) -> RuleSet:

Direct Prompt Extraction

High
Category
System Prompt Leakage
Confidence
85% confidence
Finding

Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.

Content

Scanner excerpt · core/rules.py (reported line 142)May include surrounding context.

python
raise FileNotFoundError(f"Rules file not found: {path}")
    with open(path, "r", encoding="utf-8") as f:
        data = json.load(f)
    return RuleSet(data)


def merge_rules(base: RuleSet, user_rules: Dict[str, Any]) -> RuleSet:

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding

The skill advertises file-based commands that read input files and write output files, but it does not declare any tool scope or allowed-tools constraints. In an agent environment, undeclared file read/write capability weakens least-privilege controls and can lead to unintended access to sensitive local data or overwriting user files.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The manifest description and title explicitly position the skill as a Chinese-only tool (e.g. '中文AI文本检测与改写工具' and '中文版') without stating that language selection is optional or limited to a justified region-specific context. This is a natural-language locale constraint that can violate language-choice policy when presented as the default behavior without opt-in.

Content

No source excerpt is available for this finding.

Ssd 2

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill is explicitly designed to rewrite text to evade AI-origin detection by removing chatbot markers and making machine-generated text appear human. That materially enables academic fraud, policy evasion, impersonation, and bypass of provenance or moderation systems; the context makes this more dangerous because evasion is the core advertised purpose, not an incidental side effect.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

This Python file contains multiple natural-language comments and docstrings exclusively in Chinese, including user-facing descriptive text for the detector classes. Under the policy, forcing a specific language or locale without opt-in or a documented regional justification is a natural-language policy concern.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

This Python file presents the skill's interface and behavior descriptions entirely in Chinese, including the class docstring that defines the rewriter's purpose. The policy requires flagging language or locale constraints when a skill forces a specific language without offering user choice or documenting justified regional scope.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

This Python file contains natural-language indicators that the skill is designed around Chinese locale behavior, including a default file named "zh_rules.json" and Chinese-only docstrings/comments. Under the policy rule for language/locale, forcing or assuming a specific language without explicit user opt-in can be a violation unless clearly justified as region-specific.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
78% confidence
Finding

This JSON resource consists entirely of Chinese lexical mappings and contains no natural-language indication that the skill offers a language or locale choice. Under the policy, forcing a specific language without user opt-in can be a locale-policy violation, and this file suggests a hardcoded Chinese-only behavior.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The natural-language entries include explicit cutoff phrases such as "截至2023年", "截至2024年", and "截至2025年". Embedding fixed temporal cutoffs in skill behavior can create policy issues around imposed temporal framing or stale locale/policy assumptions when no opt-in or justification is provided.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
83% confidence
Finding

This code file contains multiple natural-language comments and docstrings in Chinese, such as the section header at L009 and function docstring at L018, with no indication that the skill supports other languages or that Chinese-only operation is an intentional, justified regional constraint. Under the policy rule for language or locale, forcing a specific language without opt-in can be a violation.

Content

No source excerpt is available for this finding.

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Confidence
50% confidence
Finding

Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.

Content

Scanner excerpt · scripts/compare.py (reported line 84)May include surrounding context.

python
]
        
        for name, attr in categories:
            before_val = getattr(before, attr)
            after_val = getattr(after, attr)
            diff = after_val - before_val
            diff_str = f"+{diff}" if diff > 0 else str(diff)

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Confidence
50% confidence
Finding

Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.

Content

Scanner excerpt · scripts/compare.py (reported line 85)May include surrounding context.

python
for name, attr in categories:
            before_val = getattr(before, attr)
            after_val = getattr(after, attr)
            diff = after_val - before_val
            diff_str = f"+{diff}" if diff > 0 else str(diff)
            print(f"{name:<25} {before_val:<15} {after_val:<15} {diff_str}")

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

The docstring at L017 uses Chinese ("打印格式化检测报告") while the rest of the command-line interface strings are in English. Forcing or mixing locales without user opt-in can violate language/locale policy expectations, especially in a general-purpose CLI.

Content

No source excerpt is available for this finding.