Security audit

guaikei-douyin-competitor-watch

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly does what it claims, but its broad auto-triggering, automatic data logs, and misleading token-error behavior need human review before install.

Install only if you are comfortable sending Douyin search terms, URLs, and your Guaikei API token to www.guaikei.com and with scraped results being saved automatically under the skill's logs directory. Prefer confirming the platform before use, avoid sensitive keywords or private investigations, and review or delete generated logs after runs.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (8)

Description-Behavior Mismatch

Medium

Confidence: 88% confidence
Finding: The CLI is presented as a read/query tool for viewing Douyin posts, but it also persists the fetched results to a local JSON file. Undisclosed local storage expands the data handling surface, can leave behind sensitive or proprietary scraped content, and may violate user expectations or downstream privacy/compliance requirements.

Vague Triggers

Medium

Confidence: 93% confidence
Finding: The trigger text says the skill can activate even when the user does not explicitly mention Douyin, which makes invocation overly broad and increases the chance the agent routes unrelated requests into this skill. That can cause unintended third-party data access, surprise network calls using configured credentials, and confusion about platform scope.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The usage rules explicitly instruct use of the skill for generic short-video tasks even without mentioning Douyin, creating ambiguous activation boundaries. In context, this is more dangerous because the skill can query external data sources with an API token, so a misroute is not just a UX issue but an unnecessary credentialed action against a third-party service.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The README emphasizes scraping Douyin public data and later mentions JSON logs under the skill directory, but it does not clearly warn users up front that executing commands will persist scraped videos, author metadata, and especially comment data to local files. This can lead to unintentional retention of potentially sensitive or regulated data, increasing privacy, compliance, and accidental disclosure risk if logs are shared, synced, or left unprotected.

Missing User Warnings

Low

Confidence: 95% confidence
Finding: The CLI writes the full comment results to a local JSON file automatically after execution, but there is no clear opt-in, prior warning, or control over whether potentially sensitive scraped data is persisted. This creates a privacy and data-handling risk: comments may contain personal data or sensitive content, and unexpected local retention can expose that data to other local users, backups, or later compromise of the host.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The script writes fetched results to a local JSON file without prior disclosure or user consent. Silent persistence can leak scraped content to other local users/processes, create unexpected retention of potentially sensitive comment/post data, and materially exceeds the user's likely expectation for a lookup command.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The CLI persists full search output to a local JSON file using a filename derived from the user keyword, without any visible consent prompt, retention notice, or path disclosure. This can expose potentially sensitive search terms and returned content to other local users, backups, logs, or downstream tooling, especially on shared machines or automation environments.

Missing User Warnings

Medium

Confidence: 81% confidence
Finding: The skill transmits the user-provided keyword and search parameters to a remote API to perform the search, but the CLI help and UX do not clearly warn users that their input leaves the local environment. In a search tool, this behavior is expected functionally, but the lack of explicit disclosure creates a privacy risk if users submit sensitive queries or operationally sensitive topics.

Static analysis

No suspicious patterns detected.