The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/trends_us_list.py uses subprocess.run to execute browser-based data collection logic. It also implements dynamic module loading via importlib.util in several scripts, such as scripts/radar.py, to resolve and execute local dependencies from paths computed at runtime.\n- [REMOTE_CODE_EXECUTION]: Setup requires the installation of the Playwright Chromium browser via python -m playwright install chromium, which downloads and executes platform-specific binaries.\n- [EXTERNAL_DOWNLOADS]: Fetches analytics data and reports from trends.google.com. It also downloads browser tools from Microsoft's official Playwright registry during setup.\n- [DATA_EXFILTRATION]: Performs outgoing network requests to Google Trends endpoints and allows for user-defined proxy settings via the GOOGLE_TRENDS_PROXY environment variable.\n- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface through ingested keyword data. Evidence: Data is ingested from trends.google.com (RSS, CSV, and API responses). Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present. Capability inventory: Includes subprocess execution and external network requests. Sanitization: The skill relies on standard data parsing libraries (CSV/JSON/XML) without security-specific sanitization of keyword strings for LLM safety.

google-trends