tradingview-reader

Warn

Audited by Gen Agent Trust Hub on May 10, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses dynamic context injection (the !``command``` syntax) within the SKILL.mdfile to execute shell commands during the skill loading phase. This is used to check the local installation status ofopencli` and the TradingView plugin.
  • [EXTERNAL_DOWNLOADS]: The instructions require the user to install a global Node.js package @jackwener/opencli and a separate plugin hosted on GitHub at github.com/himself65/finance-skills.
  • [DATA_EXFILTRATION]: The skill accesses sensitive session information by retrieving cookies from the TradingView application using the Storage.getCookies command via the Chrome DevTools Protocol. While intended for authenticated data retrieval, this represents a significant data access capability.
  • [PROMPT_INJECTION]: The skill displays external data, such as news headlines and screener results, which creates a surface for indirect prompt injection. The current implementation lacks explicit boundary markers or delimiters to separate this untrusted data from the agent's instructions.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 10, 2026, 09:50 AM