Scrapling Official Skill

• Outcome: Review
• Latest audit: Aug 23, 2026 · 8:04 PM UTC
• Version: 0.4.15

Overview

This is a real scraping helper, but it gives agents broad stealth scraping, anti-bot bypass, arbitrary HTTP methods, and persistence capabilities that users should review carefully before installing.

Install only if you specifically need a powerful scraping/crawling tool and are prepared to keep its use limited to sites you own or are authorized to scrape. Avoid using the stealth, Cloudflare-solving, proxy rotation, DELETE/PUT/POST, remote CDP, persistent browser profile, and response-cache features unless you understand the legal, account, and data-retention impact. Do not use it for paywalls, authenticated areas without permission, personal data, or broad unsupervised crawling.

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 Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection

Findings (11)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 95% confidence
• Finding: The skill clearly enables outbound network access for scraping, browser automation, proxy use, and crawling, but the metadata does not declare corresponding permissions. Undeclared capabilities reduce transparency and policy enforcement, making it easier for an agent or user to invoke networked behavior without an explicit trust signal or review gate.

Description-Behavior Mismatch

Medium

• Confidence: 94% confidence
• Finding: The documentation presents the server as a scraping tool but explicitly exposes arbitrary HTTP methods including POST, PUT, and DELETE. In an MCP context, that materially expands capability from read-only retrieval to state-changing actions against arbitrary URLs, creating SSRF-style reachability and unintended side-effect risks if an agent or user assumes the tool is fetch-only.

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: The trigger text is broad enough to match many ordinary website-related requests, including cases where the user did not intend to invoke a powerful scraping or anti-bot-bypass tool. Over-broad invocation can cause the agent to escalate from simple browsing to stealth automation, proxy use, or protected-site interaction without sufficiently specific user intent.

Missing User Warnings

Low

• Confidence: 79% confidence
• Finding: The CLI workflow repeatedly instructs saving remote content to local files, including temp files, but does not prominently warn that fetched data may contain sensitive material, untrusted content, or persistent artifacts on disk. This can lead to unintended local retention of scraped data, especially in shared or multi-tenant environments.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The example explicitly promotes bypassing anti-bot protections, fingerprinting defenses, and auto-solving Cloudflare challenges without any warning about legal, ethical, or platform-policy risks. In a scraping skill, this materially increases misuse potential by normalizing evasive behavior against protected sites and may lead users to violate site access controls or trigger abuse against third-party services.

Missing User Warnings

Medium

• Confidence: 85% confidence
• Finding: The README explicitly promotes `StealthySession` for 'Cloudflare, fingerprint bypass' without any cautionary language about legal, ethical, privacy, or account/system-impact implications. In a skill specifically designed for scraping and anti-bot bypass, normalizing stealth and bypass techniques increases the chance of misuse against protected sites and makes the capability more dangerous than generic browser automation documentation.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The documentation explicitly promotes bypassing anti-bot protections and automatically solving Cloudflare challenges, but does not include clear legal, ethical, or terms-of-service warnings. In an agent skill context, this increases the chance that downstream users or autonomous agents will use the capability to access protected sites in ways the site operator did not authorize.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The example provides ready-to-use scraping code for a protected commercial site and pairs it with language about bypassing protection, which materially lowers the barrier to misuse. In a skill meant to help agents scrape websites, such examples can operationalize unauthorized data extraction against real targets without any safety framing.

Missing User Warnings

Medium

• Confidence: 82% confidence
• Finding: The response cache stores every fetched response on disk and replays it later, which can retain sensitive page content, tokens, personal data, or proprietary material from scraped sites. In a scraping tool with stealth and anti-bot bypass context, users may collect authenticated or protected content, making silent local persistence materially riskier if operators do not realize the cache is being created.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The documentation explicitly promotes anti-bot bypass and Cloudflare-solving capabilities but does not include any warning about legal, contractual, or policy restrictions. In a scraping-focused skill, this omission materially increases misuse risk because it normalizes evasion of site protections without guiding users to obtain authorization or assess compliance obligations.

Missing User Warnings

Medium

• Confidence: 88% confidence
• Finding: The example shows a hardcoded bearer token in an Authorization header without warning against embedding secrets in code, logs, or documentation. Even though the token is illustrative, this pattern can encourage insecure secret handling and accidental credential exposure when users adapt the example for real services.

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.