tradingview-reader
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_CONTEXT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of external software and plugins to function. Specifically, it instructs the installation of the
@jackwener/openclipackage from NPM and thetradingviewplugin from the author's GitHub repository (himself65/finance-skills). - [COMMAND_EXECUTION]: The skill's operation revolves around executing shell commands via the
opencliutility. This includes commands to query market data, capture screenshots, and relaunch the TradingView desktop application with specific debugging flags (--remote-debugging-port=9222). - [DYNAMIC_CONTEXT_INJECTION]: The
SKILL.mdfile contains a shell command executed at skill load time using the!commandsyntax. This command ((command -v opencli && opencli tradingview status 2>&1 | head -5 && echo "READY" || echo "SETUP_NEEDED") 2>/dev/null || echo "NOT_INSTALLED"``) is used to detect the environment's readiness and the presence of required tools. - [INDIRECT_PROMPT_INJECTION]: The skill ingests market data, news stories, and user watchlist content from TradingView's APIs. This creates an attack surface where malicious content embedded in those sources (e.g., a specially crafted news headline) could potentially influence the agent's behavior.
- Ingestion points: Data is fetched via
opencli tradingviewcommands from TradingView's scanner, news, and alert APIs. - Boundary markers: The instructions suggest using structured formats like JSON/YAML for output but do not include explicit instructions to ignore embedded commands within the fetched content.
- Capability inventory: The agent can execute shell commands and read local files via the provided toolset.
- Sanitization: There is no evidence of sanitization or filtering of the external API content before it is processed by the agent.
- [DATA_EXPOSURE]: The skill documentation explicitly states that the adapter harvests session cookies via the Chrome DevTools Protocol (
Storage.getCookies) to replicate the user's authenticated session for API requests. While this is a functional requirement for reading account-specific data, it involves the programmatic access of sensitive authentication tokens.
Audit Metadata