Ai Compliance Review

• Outcome: Review
• Latest audit: Jul 19, 2026 · 6:56 AM UTC
• Version: 1.1.0

Overview

This compliance-review skill does what it claims, but it can send user-selected files and whole directories of potentially sensitive documents to an external DeepSeek-compatible API without a clear per-run disclosure or consent step.

Use this skill only for documents you are allowed to send to a DeepSeek-compatible external API. Avoid running batch or recursive scans over directories that may contain confidential contracts, HR files, customer data, credentials, exports, JSON, or CSV data unless you have reviewed the file set first.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (5)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 94% confidence
• Finding: The skill declares executable capabilities in practice (environment variable use, file read/write, and outbound network access to DeepSeek) but does not declare corresponding permissions. This weakens user awareness and platform enforcement, and in this skill’s context it is especially relevant because reviewed documents may contain sensitive legal, HR, financial, or privacy-related content that can be transmitted externally and written to disk.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The skill requires an external API key and documents a remote base URL, but it does not clearly warn users that submitted text or file contents may be sent to a third-party API service for analysis. In this skill’s context, inputs are likely to include contracts, privacy policies, marketing drafts, employee documents, and other sensitive materials, creating meaningful confidentiality, privacy, and compliance risk.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The tool reads arbitrary local file contents and sends them to an external DeepSeek API for analysis without an explicit just-in-time warning or consent check. Users may provide contracts, privacy policies, employee data, or other sensitive material, creating confidentiality and regulatory exposure if they do not realize the content leaves the local environment.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The batch scan mode can transmit the contents of every matched file in a directory, potentially including many sensitive or unrelated documents, with no per-run disclosure or confirmation. This greatly amplifies accidental data leakage risk because users may not appreciate the breadth of files included, especially with recursive scanning.

External Transmission

Medium

• Category: Data Exfiltration

Content

}

    try:
        resp = requests.post(url, headers=headers, json=data, timeout=120)
        if resp.status_code >= 400:
            print(f"API请求失败 (HTTP {resp.status_code}): {resp.text}", file=sys.stderr)
            sys.exit(1)

• Confidence: 96% confidence
• Finding: requests.post(url, headers=headers, json=

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

Detected: suspicious.exposed_secret_literal

File appears to expose a hardcoded API secret or token.

Critical

• Code: suspicious.exposed_secret_literal
• Location: SKILL.md:55