captcha-bypass

Fail

Audited by Snyk on Mar 11, 2026

Risk Level: HIGH
Full Analysis

HIGH W007: Insecure credential handling detected in skill instructions.

  • Insecure credential handling detected (high risk: 1.00). The prompt includes code/config examples that require inserting API keys and proxy credentials (e.g., CAPTCHA_API_KEY, capsolver.api_key, proxy username/password) directly into requests or scripts, which would force the LLM to handle/output secret values verbatim.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly requires using a browser tool to navigate Cloudflare/target websites and to submit site URLs/site-keys to third-party CAPTCHA solvers (e.g., the "Step 2: Use the Browser Tool" section and the 2Captcha/CapSolver examples), which means the agent will fetch and interpret arbitrary public web content that can influence its subsequent actions.
Audit Metadata
Risk Level
HIGH
Analyzed
Mar 11, 2026, 01:25 AM
Security Audit — snyk — captcha-bypass