superme

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill orchestrates browser automation via the browser-use CLI, employing eval commands to execute JavaScript within the web context. This is used to interact with the DOM, manipulate Vue.js and AngularJS application states, and perform automated logins.
  • [DATA_EXFILTRATION]: Sensitive session data, including authentication tokens, JWTs, and browser cookies, are exported and saved to the local filesystem (e.g., /tmp/superme_cookies.json, /tmp/superme_rami_token.txt). While required for session persistence across commands, this represents local exposure of session credentials.
  • [EXTERNAL_DOWNLOADS]: The skill automatically installs the browser-use tool using the uv package manager if it is not detected in the environment during execution.
  • [PROMPT_INJECTION]: The skill's documentation includes specific instructions for AI agents regarding safe installation procedures to mitigate a known supply chain attack in its dependencies. These instructions are directed at the agent's execution logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 07:22 PM