ultimate-browsing

Warn

Audited by Gen Agent Trust Hub on Aug 9, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The script scripts/extract_cookies.py is designed to extract highly sensitive authentication cookies from local browser profiles (Chrome, Firefox, Brave, etc.). It programmatically accesses and decrypts data from OS-level credential managers, including the macOS Keychain, Linux SecretService, and Windows DPAPI.
  • [COMMAND_EXECUTION]: The skill utilizes subprocess.run to execute local Node.js templates in engine/executor.py. Additionally, scripts/extract_cookies.py executes system commands to retrieve passwords and runs dynamic JavaScript code using the node -e flag.
  • [EXTERNAL_DOWNLOADS]: The skill instructions direct the agent to install several external dependencies and binaries at runtime, including cloakbrowser (from CloakHQ), agent-browser (from Vercel Labs), and various Python/Node.js scraping libraries.
  • [PROMPT_INJECTION]: The references/insane-search/README.md file contains 'Harness Rules' (R1-R7) that function as mandatory behavioral overrides, instructing the agent to follow specific failure-handling and escalation logic that may conflict with default safety or operational constraints.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 9, 2026, 12:43 PM
Security Audit — agent-trust-hub — ultimate-browsing