guaikei-douyin-scout

• Outcome: Review
• Latest audit: Aug 24, 2026 · 3:44 AM UTC
• Version: 1.0.0

Overview

This Douyin public-data skill is mostly coherent, but it needs Review because it can run on ambiguous research prompts and automatically saves fetched data locally.

Install only if you are comfortable sending Douyin queries, creator/video URLs, and the GUAIKEI_API_TOKEN to guaikei.com, and with search/post/comment results being automatically saved under the skill's logs directory. Use explicit Douyin prompts, avoid sensitive targets or personal-data profiling, periodically delete logs, and treat the hardcoded WeChat support message as an unverified private contact path.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (13)

Description-Behavior Mismatch

Medium

• Confidence: 91% confidence
• Finding: The changelog documents capabilities beyond the skill metadata’s stated four Douyin use cases, including broader social-data search and analytics. This creates a scope mismatch that can enable undisclosed data collection or unauthorized feature exposure, making review, consent, and policy enforcement harder.

Context-Inappropriate Capability

Medium

• Confidence: 93% confidence
• Finding: The documentation explicitly mentions competitor-account bulk scraping and social-data analytics, which exceed the user-facing purpose of simple search, creator posts, comments, and trending topics. Even in a Douyin-focused skill, this context makes the issue more concerning because it suggests scaling from normal retrieval into surveillance-style collection and potentially non-compliant scraping behavior.

Description-Behavior Mismatch

Low

• Confidence: 89% confidence
• Finding: The CLI persists the full comment retrieval output to a local JSON file after displaying it, which is a data-handling behavior not implied by the stated user-facing purpose of simply fetching comments. This creates an unnecessary retention surface for potentially sensitive or user-generated content on disk, where it may later be accessed by other users, processes, backups, or log collectors.

Description-Behavior Mismatch

Medium

• Confidence: 91% confidence
• Finding: The CLI persists full search results to a local file without any visible consent, retention policy, or minimization. Search results may contain user-generated content, metadata, or sensitive query context, so silent persistence expands the data exposure surface beyond the stated browse/search behavior.

Context-Inappropriate Capability

Low

• Confidence: 92% confidence
• Finding: The authentication error message embeds an out-of-band personal WeChat contact and instructs users to use it to obtain a solution. This creates an unnecessary social-engineering and trust boundary issue: users encountering auth failures may be pushed to contact an unverified private channel, which can enable phishing, unauthorized support flows, or privacy leakage. In a Douyin-related skill that depends on external APIs, this is more suspicious because users may assume the contact is officially required to restore service.

Vague Triggers

High

• Confidence: 95% confidence
• Finding: The activation guidance says the skill should trigger even when the user does not explicitly mention Douyin, including broad research requests embedded in multi-step workflows. That can cause the agent to invoke this skill on generic research tasks, sending user queries to an external service unexpectedly and broadening data exposure beyond the user's clear intent.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: Several trigger phrases such as '网上火啥', '评论区啥风向', and generic search-style wording overlap with ordinary conversation and are not uniquely scoped to Douyin. This increases the chance of accidental invocation, causing unintended external requests and potentially logging user prompts or targets without sufficiently specific consent.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill states that outputs are automatically saved to `logs/` with filenames containing timestamps and query parameters, but it does not prominently warn users that their search terms and target URLs may be written to disk. Query terms may contain sensitive research topics, personal handles, or investigation targets, so silent persistence increases privacy and retention risk.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The README states that data export and automatic JSON log generation occur, but it does not present this as a prominent privacy/compliance warning before use. Because the skill scrapes public platform data and stores it locally by default, users may unintentionally retain potentially sensitive profile, comment, or trend data without understanding storage, sharing, or retention implications.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The skill promotes comment analysis and competitor-account monitoring but only includes a brief restriction against improper redistribution, without a clear privacy/compliance warning about collecting, profiling, or analyzing third-party public content. This increases the risk of misuse for surveillance, improper personal-data handling, or terms-of-service and regulatory violations, especially in marketing and monitoring contexts.

Missing User Warnings

Medium

• Confidence: 78% confidence
• Finding: The tool makes external API requests and later writes retrieved data locally without meaningful user-facing disclosure beyond generic status messages. In a skill context, hidden network transmission and undisclosed local persistence reduce informed consent and can surprise users about where their data goes and what artifacts remain on the system.

Missing User Warnings

Medium

• Confidence: 84% confidence
• Finding: The CLI writes the fetched Douyin results to a local JSON file automatically, without any consent prompt, opt-in flag, or visibility about retention. Because the data may include scraped creator content and metadata, this can create unintended local persistence and exposure to other local users, backup systems, or later compromise of the host.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: Writing search results to disk without explicit user warning or consent creates a privacy and data-retention risk, especially for a user-facing search/browse skill where users may reasonably expect transient processing only. If the host environment is shared or compromised, these log files could expose query history and returned content to unintended parties.

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.