Scrapling Official Skill

• Outcome: Review
• Latest audit: Jul 12, 2026 · 8:00 PM UTC
• Version: 0.4.11

Overview

This appears to be a legitimate Scrapling scraping skill, but it gives agents powerful anti-bot, proxy, and persistent-browser capabilities with uneven guardrails.

Install only if you need advanced web scraping. Use it only on sites you are authorized to access, keep robots.txt/ToS and rate limits in force, avoid paywalls or authenticated areas without permission, close persistent sessions, and clear any local crawl or development caches that may contain scraped content.

SkillSpector

By NVIDIA

Vulnerability Patterns

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

Findings (11)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 88% confidence
• Finding: The skill clearly enables outbound network access through HTTP requests, browser automation, crawling, and proxy use, yet the metadata does not declare permissions. Missing permission declarations can prevent proper policy enforcement and user awareness, increasing the chance the skill is invoked in contexts where networked behavior is unexpected or disallowed.

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The trigger text is very broad: it activates for generic scraping, crawling, extraction, failed web_fetch, anti-bot sites, and even writing Python code. Overbroad routing can cause the skill to be selected for loosely related user requests, unnecessarily expanding access to networked scraping and stealth features and increasing the chance of misuse or policy bypass.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The skill prominently advertises stealth browsing, Cloudflare/anti-bot bypass, proxy rotation, and challenge solving, but the warnings are limited and do not clearly frame these as high-risk actions requiring explicit authorization. In context, this makes the skill more dangerous because its core purpose includes techniques commonly used to evade site controls, which can facilitate unauthorized access patterns, privacy violations, or terms-of-service abuse.

Missing User Warnings

Medium

• Confidence: 88% confidence
• Finding: The example explicitly promotes stealth browsing and automatic anti-bot bypass capabilities, including Cloudflare challenge solving, but does not provide any warning about authorization, terms-of-service, rate limits, or legal/policy constraints. In a scraping-focused skill, normalizing covert access techniques without guardrails increases the risk that users will deploy the code against protected sites in ways that violate access controls or trigger abuse-related impact.

Missing User Warnings

Medium

• Confidence: 77% confidence
• Finding: The document promotes stealthy headers, browser impersonation, proxy support, and form submission but does not pair those capabilities with clear warnings about authorization, privacy, credential handling, or acceptable use. In a scraping-focused skill, omission of such guardrails can normalize evasive or sensitive-data-transmitting behavior and increase the chance of unauthorized collection or misuse.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The documentation explicitly promotes stealth scraping, anti-bot evasion, fingerprint masking, and automatic challenge solving without any compliance, authorization, or rate-limit guidance. In a scraping-focused skill, this materially lowers barriers to misuse against protected sites and normalizes bypassing defensive controls, increasing the risk of unauthorized collection, ToS violations, and service disruption.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The examples provide ready-to-use code for automatic Cloudflare challenge solving, stealth options, and proxy use, but omit any caution about authorization, legality, or operational impact. Concrete examples are especially dangerous because they are directly reusable, enabling users to target protected websites with minimal effort and making abusive use more likely.

Missing User Warnings

Medium

• Confidence: 85% confidence
• Finding: Persistent sessions are documented as reusable across calls, and the docs mention cookies and browser state parameters without warning that authentication state, cookies, and browsing context may persist and be reused unintentionally. In an agent setting, this increases the risk of cross-task data exposure or unintended actions performed with retained authenticated state, especially because `list_sessions` and session reuse are first-class features.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The documentation presents `robots_txt_obey = False` as a configurable/default behavior without clearly warning that this bypasses site-declared crawl restrictions and can lead users to perform scraping that violates operator policies or triggers blocking and abuse responses. In a scraping skill explicitly advertising anti-bot bypass and stealth browsing, normalizing robots.txt non-compliance materially increases the likelihood of misuse.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The documentation explicitly promotes anti-bot bypass and Cloudflare-solving capabilities, including examples of `AsyncStealthySession` and `solve_cloudflare=True`, without any warning about authorization, terms-of-service, or legal/compliance boundaries. In a scraping-focused skill, this materially increases misuse risk because it normalizes evasion of access controls rather than framing these features for permitted testing or owned environments only.

Unsafe Defaults

Medium

• Category: Tool Misuse

Content

class MySpider(Spider):
    name = "my_spider"
    start_urls = ["https://example.com"]
    development_mode = True

    async def parse(self, response: Response):
        yield {"title": response.css("title::text").get("")}

• Confidence: 88% confidence
• Finding: development_mode = True

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.