guaikei·小红书内容与博主数据

• Outcome: Pass
• Latest audit: Aug 16, 2026 · 1:54 AM UTC
• Version: 1.0.0

Overview

This is a coherent Xiaohongshu public-data research tool, but it sends user queries and links to guaikei and saves returned results locally.

Install only if you are comfortable sending Xiaohongshu keywords, note URLs, profile URLs, and request metadata to guaikei and storing returned public data in local JSON logs. Avoid using confidential research terms or sensitive/tokenized links on shared machines, and periodically delete or protect the logs directory if the outputs contain business research or personal data.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• 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 (7)

Description-Behavior Mismatch

Low

• Confidence: 88% confidence
• Finding: The skill states that every execution result is automatically archived to a local logs/ directory, which creates persistent storage of queried URLs, keywords, and returned data. Even if the source data is public, local retention expands exposure through disk access, backups, or accidental sharing, especially when user-supplied links may contain tracking or tokenized query parameters.

Intent-Code Divergence

Low

• Confidence: 84% confidence
• Finding: The documentation presents the skill as a retrieval-only integration, but elsewhere describes automatic writing of results to disk. That mismatch can mislead users and downstream agents about side effects, causing them to provide data under the false assumption that nothing is retained locally.

Missing User Warnings

Low

• Confidence: 86% confidence
• Finding: The skill depends on a third-party API service, but the disclosure that user-provided keywords and links are sent off-box appears only later in a compliance note. Insufficient upfront notice can lead users to share internal research terms or sensitive URLs without realizing they will be transmitted to an external provider.

Missing User Warnings

Medium

• Confidence: 86% confidence
• Finding: The README states that all task results are automatically saved to a local logs/ directory, but it does not warn users that these results may include queried keywords, note URLs, profile links, comment data, or derived analysis outputs. In a data-scraping and competitor-analysis tool, such logs can accumulate sensitive business intelligence or personal data and may be exposed through shared workstations, backups, version control, or insecure retention practices.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The CLI writes retrieved comment results to a local JSON file automatically without clearly warning the user in the help text or offering an opt-in/opt-out control. Because comment data may contain personal data, sensitive business research, or regulated content, silent persistence increases the risk of unintended disclosure through shared workstations, backups, artifact collection, or later exfiltration.

Missing User Warnings

Low

• Confidence: 81% confidence
• Finding: The tool persists fetched results to a local JSON file automatically, but the CLI help and runtime output do not clearly warn users that note details and comment data will be written to disk. This can create unintended local data retention and exposure risks on shared systems, especially because the saved content may include substantial scraped public data and potentially sensitive research artifacts.

Missing User Warnings

Medium

• Confidence: 87% confidence
• Finding: The CLI persists the full structured output, including the user-supplied profile URL and fetched creator/post data, to disk via log.taskWrite without any consent prompt, data-minimization, or masking. Even though the skill is intended for public Xiaohongshu data, storing raw URLs and results can create unnecessary privacy and compliance exposure, especially if URLs contain tracking/query tokens such as xsec_token or if logs are retained/shared insecurely.

VirusTotal

64/64 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.