Security audit

Douyin Scraper

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed Douyin browser-scraping skill, but its very broad auto-activation and weak handling guidance for saved login state warrant review before installation.

Install only if you intentionally want browser-based Douyin searching. Use explicit Douyin wording when invoking it, avoid sensitive search terms, do not rely on vague topic prompts, and treat any saved douyin-auth.json file like a password: keep it out of source control, do not share it, and delete it when no longer needed.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (13)

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The README advertises very broad natural-language triggers such as generic search requests, which can overlap with ordinary user conversation and cause accidental skill activation. In an agent environment, this increases the chance of unintended browser automation against Douyin, potentially launching searches and scraping actions without explicit user intent.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The usage section explicitly encourages unconstrained free-form triggering ('就这么简单') without any activation guardrails. In a multi-skill or autonomous agent setting, this makes prompt-based misrouting and accidental execution more likely, especially because the skill performs browser automation and data extraction rather than a harmless local transformation.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The supported utterances include extremely generic phrases like '健身视频' with no negative examples or disambiguation guidance. Because the skill context is browser-driven scraping, overly short trigger phrases materially increase the risk of accidental or adversarial invocation from normal dialogue content.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The README instructs users to save persistent login state to a local file (douyin-auth.json) but does not warn that this file may contain reusable authentication tokens, cookies, or session data. If the file is exposed through logs, backups, source control, or other local compromise, an attacker may hijack the authenticated Douyin session and access account-bound content.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The phrase '想说什么就说什么，Skill 会自动理解并执行' makes activation effectively unconstrained and suggests the skill may trigger on arbitrary conversational text rather than explicit scraping intent. In an agent setting, overly broad triggers can cause unintended browser automation and data collection when a user says something that loosely resembles a search request.

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The listed trigger phrases are common everyday requests ('帮我搜…', '找一下…', '搜索…') and are not uniquely tied to this skill. Because this skill performs browser-driven scraping on a third-party site, broad natural-language matching increases the chance of accidental activation and unintended navigation or collection of content.

Missing User Warnings

Medium

Confidence: 87% confidence
Finding: The skill advertises persistent browser state and later documents saving and loading authentication state, but does not warn that cookies, tokens, and browsing artifacts may be retained locally. This creates privacy and account security risk if stored session files are reused, exposed, or shared, especially for a scraper interacting with a logged-in social media account.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The description includes a very broad natural-language trigger example ("说'搜索一下海鲜视频'即可执行搜索"), which can overlap with ordinary user conversation and cause the skill to activate unintentionally. In a browser-automation scraper, accidental activation is more dangerous because it can launch navigation and scraping actions without clear user intent, creating privacy, consent, and unwanted-action risks.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: These trigger examples are broad natural-language phrases that can easily overlap with ordinary user conversation, making the skill prone to accidental activation. In a browser-automation scraping skill, unintended activation can launch searches or automated browsing without the user explicitly invoking the skill, increasing the risk of surprise actions and data exposure.

Vague Triggers

High

Confidence: 97% confidence
Finding: The minimal triggers are extremely ambiguous single words or short phrases like topic names, which provide almost no boundary between casual conversation and a command to execute the skill. Because this skill performs browser automation against Douyin, such ambiguous activation can cause unintended searches, navigation, and scraping behavior from innocuous user text.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The colloquial examples mirror normal conversational requests and do not define where casual discussion ends and tool execution begins. This is risky in agent environments because routine dialogue may be misclassified as a command, leading the browser automation skill to perform actions the user did not intentionally authorize.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The script makes an external network request to Douyin and writes a screenshot to /tmp using user-influenced input in the filename, but it provides no prior consent prompt or warning about outbound access and local artifact creation. In an agent/skill context, this can unexpectedly leak user intent to a third-party service and leave recoverable local files on disk, which is a real security and privacy issue even if the functionality is intentional.

Vague Triggers

High

Confidence: 95% confidence
Finding: The activation patterns are broad enough to trigger on generic requests like '搜一下' or '搜索.*视频', which can cause this skill to hijack unrelated user intents. Because the skill performs browser-based scraping, accidental invocation expands the attack surface and may lead to unintended browsing, data access, or policy-violating automation.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.