captcha

This module is a straightforward CAPTCHA-solving client that communicates with fixed third-party solver endpoints, sending the API key and challenge data (and optionally base64-encoded local image contents) and polling until a solution token is returned. No overt malware techniques (backdoors, persistence, command execution, or obfuscation) are evident. The main security concern is abuse-enabling functionality plus data disclosure to external services, including API-key exposure in a polling GET query string and unrestricted local file reading in image mode (path provided by the caller). Overall: likely an integration/automation utility with moderate security/policy risk depending on the containing application’s intent and data-handling requirements.

High-risk supply-chain content: this module provides concrete guidance and automation steps to evade bot-detection/CAPTCHA/verification mechanisms using stealth fingerprint patching, request/header spoofing, human-like interaction simulation, persistent session warmup, and proxy routing. While it does not show traditional malware (no exfiltration/persistence/command execution in this fragment), it directly supports circumvention of access controls and verification workflows, making it unsuitable for benign use without strong justification and review.

This module is best characterized as CAPTCHA-bypass automation: it scrapes reCAPTCHA image-grid challenges from a live browser session, exfiltrates the challenge image and task text to external vision LLM APIs, uses the model output to select and click tiles/verify, and includes basic automation-evasion (navigator.webdriver modification). While there is no clear evidence of system compromise or credential-stealing malware in this file, its intended/operational behavior is high-risk from a security-policy and abuse perspective, especially when introduced via a supply chain without explicit authorization and user consent.

SUSPICIOUS. The skill is internally consistent with its stated purpose, but that purpose is to bypass anti-bot/security controls and route challenge data to third-party solvers so automation can proceed. There is no strong evidence of hidden malware or deceptive install behavior, yet the combination of CAPTCHA bypass, external data transfer, stealth guidance, and automated continuation of protected actions makes it a high-risk AI agent capability.

No direct signs of classic malware are evident in the provided fragment; it does not show system-compromising payloads. However, the described capability is explicitly intended to bypass CAPTCHA and anti-bot protections by obtaining tokens from third-party services, injecting them into browser pages, and invoking internal callbacks (plus optional LLM-assisted visual solving and evasion guidance). This is a high-risk misuse/abuse enabler in a supply-chain context, warranting strict review, policy controls, and limitations on distribution and use.

This module provides an end-to-end playbook to bypass CAPTCHA-based bot defenses: it detects CAPTCHA type by inspecting the DOM/page source, extracts provider parameters (sitekeys/callback hints), obtains solved tokens/answers via external solving workflows (including image capture/download), injects solutions into the corresponding hidden response inputs, and triggers or subverts provider verification (including overriding reCAPTCHA v3 execution). While the snippet does not show classic payload behaviors like exfiltration or persistence, it is highly aligned with security-control circumvention and should be treated as high risk for misuse if present in a dependency.