GDPR-Compliance（GDPR合规工具）

• Outcome: Pass
• Latest audit: Jul 9, 2026 · 8:18 PM UTC
• Version: 1.1.0

Overview

This appears to be a local GDPR report-generation toolkit, with no evidence of network exfiltration or destructive behavior, though users should handle generated reports carefully.

Install and run it in a virtual environment, review generated reports before sharing them, and avoid entering unnecessary personal or confidential business data. Treat the outputs as compliance aids rather than legal advice, and note that the security-check script runs local helper commands rather than only scanning files.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Behavioral AST: exec() Call, eval() Call, Dynamic Import
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection

Findings (9)

subprocess module call

Medium

• Category: Dangerous Code Execution

Content

for cmd in test_commands:
            try:
                result = subprocess.run(
                    cmd,
                    capture_output=True,
                    text=True,

• Confidence: 91% confidence
• Finding: The script invokes other project scripts with subprocess as part of a 'security check'. Even though the command list is hardcoded and shell injection is not evident, executing repository code during a safety-verification step can trigger arbitrary side effects from those target scripts, undermining the claim that this checker only verifies local/security properties.

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 84% confidence
• Finding: The skill declares itself as 'pure local' and claims no network/system-modifying behavior, yet the content advertises execution of Python scripts, file reads/writes, a cross-directory script invocation, and static analysis detected network/shell capabilities. This mismatch is dangerous because users and orchestration systems may grant trust based on the documented safety claims while the skill may exercise broader capabilities than disclosed.

Intent-Code Divergence

Medium

• Confidence: 89% confidence
• Finding: The guide presents inconsistent dependency information: the skill metadata says it relies on Python standard library plus pandas, while this document also lists jinja2 as a required package. In a security/install guide, inaccurate dependency disclosure can mislead reviewers, weaken supply-chain review, and cause users to approve or install packages they did not expect.

Context-Inappropriate Capability

Medium

• Confidence: 95% confidence
• Finding: A security-check utility should not need to execute other project programs to verify whether the project is local-only or safe. This creates an unnecessary execution capability: a malicious or compromised checked script could perform destructive local actions when the checker runs, causing the checker itself to become an execution vector.

Intent-Code Divergence

Medium

• Confidence: 84% confidence
• Finding: The docstring says the script verifies security and pure local-running properties, but the implementation also runs other scripts. That mismatch is dangerous because users may trust the tool as a passive analyzer and run it in sensitive environments without realizing it actively executes project code.

Vague Triggers

Medium

• Confidence: 80% confidence
• Finding: The trigger keywords are broad terms like GDPR, data protection, DPIA, and cross-border transfer, with no strong activation boundaries. That can cause the skill to activate in unrelated or mixed-context conversations, increasing the chance that file-writing or script-execution behavior is invoked unexpectedly on sensitive compliance content.

Vague Triggers

Medium

• Confidence: 81% confidence
• Finding: A second broad trigger list repeats expansive legal/privacy phrases without exclusions, compounding the overbroad activation risk. In a compliance skill that can run scripts and write reports, repeated generic triggers make accidental or premature invocation more likely, especially around sensitive regulatory or business documents.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The interactive mode prompts the user to enter free-form compliance notes and later saves the full results structure to disk as JSON, but the workflow does not clearly warn users that their responses will be persisted. In a GDPR/compliance context, those notes may contain sensitive internal process details, personal data, or incident-related information, creating avoidable confidentiality and data minimization risks if stored unexpectedly or insecurely.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The script executes other project scripts without a clear warning or confirmation at the execution point. In a security-check context, that lack of disclosure materially increases risk because users may believe they are performing passive analysis while actually launching untrusted code from the repository.

VirusTotal

65/65 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.