Xiaohongshu Search Summarizer

• Outcome: Review
• Latest audit: May 29, 2026 at 9:23 PM UTC
• Version:
  1.0.3

Overview

The skill largely matches its advertised Xiaohongshu research purpose, but it needs review because it scrapes and stores third-party content locally and has an output-path containment bug.

Review before installing. Use a dedicated output directory, avoid search keywords containing slashes or '..', supervise any Xiaohongshu login challenge, delete raw outputs when finished, and treat scraped comments/images as third-party material subject to privacy, copyright, and platform rules.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (6)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 96% confidence
• Finding: The skill requires and instructs use of file read, file write, network access, and likely environment-dependent tooling, but it does not declare permissions. Hidden or undeclared capabilities reduce informed consent and make it harder for a caller or platform to enforce least privilege, especially because the workflow downloads external content and writes multiple local artifacts.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 89% confidence
• Finding: The documented purpose emphasizes search and synthesis, but the actual behavior also downloads remote images and produces intermediate raw-data artifacts on disk. This mismatch can mislead users and orchestrators about what data is collected and persisted, increasing the risk of unexpected storage of third-party content and comments.

Vague Triggers

Medium

• Confidence: 79% confidence
• Finding: The invocation language is broad enough to trigger on generic research, summarization, or social-media analysis requests, not just explicit Xiaohongshu tasks. Over-broad triggering can cause unintended browsing, scraping, and local file creation in contexts where the user did not specifically request this source or this level of data collection.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The skill directs the agent to save downloaded images and a synthesized report to local storage without a clear user-facing warning. Silent file creation can surprise users, consume disk space, and persist potentially sensitive or copyrighted third-party content beyond the immediate session.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The skill collects and downloads Xiaohongshu posts, images, and user comments without any explicit privacy or data-handling warning. Because comments and media may contain personal data or sensitive content, the absence of collection, retention, and usage disclosures creates privacy and compliance risk.

Missing User Warnings

Medium

• Confidence: 87% confidence
• Finding: The script saves scraped Xiaohongshu content, including post text, image URLs, and user comments, directly to disk in the specified output directory without any consent check, minimization, retention control, or warning about handling third-party personal data. In this skill’s context, comment text and usernames can contain personal or sensitive information, so persistent local storage increases privacy, compliance, and unintended disclosure risk if the output directory is shared or reused.

VirusTotal

66/66 vendors flagged this skill as clean.