browsing

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The codebase contains deliberate anti-detection / WAF-evasion tooling plus explicit local-cookie extraction/decryption and cookie-injection flows (including listed auth cookie names and bridging them into request pools), which enable credential theft and stealthy remote access — high risk for abuse/exfiltration.

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

• Third-party content exposure detected (high risk: 0.85). Yes: the required runtime path “Tier 1/1.5/2 fetches a user-provided URL/article/feed → extracts readable HTML/text → engine wraps it as [BEGIN/END UNTRUSTED WEB CONTENT] and returns it to the agent” means outsider-authored free text from public web content is ingested into the LLM context via the browsing fetch chain.