Back to skill

Security audit

Adlaw Checker

Security checks for vulnerabilities and agentic risk

Overview

This is a straightforward Chinese advertising-copy checker with some dependency and accuracy caveats, but no evidence of hidden or malicious behavior.

Install only in an isolated environment and do not rely on its output as legal advice. The checker appears local and benign, but users should treat it as a rough keyword tool, remove or pin the unnecessary dependency install step, and manually review important advertising copy.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (2)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:29
Finding

Unpinned and Unnecessary Third-Party Dependencies in Skill Installation Instructions

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:29
Vulnerability Type: T08: Insecure Dependencies
Risk Level: Medium

Vulnerable Code Snippet:

bash
pip install jieba pandas

Technical Analysis

The installation command retrieves mutable, unpinned packages from the Python package index configured in the user's environment. It provides no version constraints, integrity hashes, lock file, or trusted-index enforcement.

Neither jieba nor pandas is imported or otherwise used by scripts/checker.py. The instruction therefore introduces avoidable supply-chain exposure without supporting the implemented functionality.

Python package installation may execute package build or installation logic. If a listed package, one of its transitive dependencies, or the configured package index is compromised, attacker-controlled code could execute during installation.

Attack Path

  1. An attacker compromises a package release, transitive dependency, package-index account, or package source selected by the user's pip configuration.
  2. The attacker publishes a malicious distribution under a version satisfying the unrestricted command.
  3. A user follows the documented installation instruction.
  4. Pip resolves and downloads the attacker-controlled distribution.
  5. Malicious build or installation logic executes with the privileges of the user running pip.

Impact Assessment

Successful exploitation could execute arbitrary code with the installing user's privileges. Depending on the environment, this may allow access to files, environment variables, credentials, project data, and network resources available to that user. If installation is performed with elevated privileges, the impact may extend to system-wide modification.

The project itself does not contain evidence of a malicious dependency. Exploitation requires compromise or manipulation of the external package supply chain or configured package ...[truncated 7 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove the installation command because the current implementation does not use either dependency.
  2. If these packages become necessary, pin reviewed versions in a dependency lock file.
  3. Require package hashes, for example through a hash-locked requirements file and pip install --require-hashes.
  4. Use an explicitly trusted package index or an internally controlled package mirror.
  5. Audit direct and transitive dependencies before updates.
  6. Install dependencies inside an isolated, least-privileged virtual environment rather than with administrative privileges.

T08 · Insecure Dependencies

Warning
Location
README.zh-CN.md:10
Finding

Unpinned and Unnecessary Third-Party Dependencies in README Installation Instructions

Content
View full analysis

Vulnerability Details

File Location: README.zh-CN.md:10
Vulnerability Type: T08: Insecure Dependencies
Risk Level: Medium

Vulnerable Code Snippet:

bash
pip install jieba pandas

Technical Analysis

The README directs users to install two unrestricted third-party packages without version pins, integrity hashes, a lock file, or package-source restrictions. Package resolution consequently depends on mutable external releases and the user's pip index configuration.

The audited implementation in scripts/checker.py does not import or use either package. This creates unnecessary exposure to package compromise, malicious transitive dependencies, and manipulated package sources.

Because Python package installation can invoke build and installation logic, a malicious distribution selected during dependency resolution may execute code before the checker is run.

Attack Path

  1. An attacker gains control of a listed package, a transitive dependency, or a package source used by the victim.
  2. A malicious, unrestricted package version is made available.
  3. The victim follows the README and executes the documented pip command.
  4. Pip retrieves and processes the malicious distribution.
  5. Attacker-controlled installation code runs under the victim's account.

Impact Assessment

The attacker could obtain code execution with the privileges and environmental access of the user performing installation. Accessible project files, user files, environment variables, development credentials, and network services may be exposed or modified. System-wide impact is possible if the command is run in an elevated context.

No evidence shows that the currently named packages are malicious. This finding concerns the avoidable supply-chain risk created by unrestricted and unused dependencies.

Remediation
View remediation

Remediation Suggestions

  1. Delete the dependency installation instruction because the current checker uses only Python standard-library modules.
  2. Keep installation documentation synchronized with actual runtime imports.
  3. If dependencies are introduced later, pin exact reviewed versions and transitive dependencies.
  4. Record cryptographic hashes and enforce them during installation.
  5. Retrieve packages only from a trusted, explicitly configured index or controlled mirror.
  6. Perform dependency installation in an isolated virtual environment with least privilege.
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The natural-language content of the skill forces a specific language/locale for users, which matches the policy-violation category for language or locale constraints without user opt-in. Although the subject matter is China-specific advertising law, the file does not explicitly state that the skill is limited to Chinese-language or China-region use as a documented justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This code initializes the checker with lang: str = "zh", and the surrounding docstrings, messages, and content are all Chinese-language specific. Under the policy, forcing a specific language or locale without user opt-in is a natural-language policy violation unless the locale restriction is explicitly justified or the user is offered a choice.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The code documents and exposes platform-specific checks via CLI values like 'xiaohongshu' and 'wechat', but the lookup uses if platform and platform in self.violation_words while the actual keys are platform_xiaohongshu and platform_wechat. As a result, valid user-supplied platform values silently skip those extra checks, causing false negatives and allowing non-compliant content to be reported as safer than it is.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

该文件整体内容仅提供中文说明,文件名也明确限定为 zh-CN,本文件内未说明这是可选本地化版本,也未向用户提供语言选择或其他语言入口。根据规则,若技能强制特定语言而无用户选择,属于自然语言层面的语言/地区策略风险。

Content

No source excerpt is available for this finding.