portfolio-health-check
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes data-fetching commands via
python ../findata-toolkit-cn/scripts/views_runner.py, relying on scripts from the author's broader toolkit. - [EXTERNAL_DOWNLOADS]: The skill's setup instructions include installing third-party Python packages from a
requirements.txtfile located in a sibling directory. - [REMOTE_CODE_EXECUTION]: The skill extends its functionality by executing Python scripts from a path external to its own directory structure (
../findata-toolkit-cn/). - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection where untrusted user input from the portfolio holding list is interpolated into shell commands. Without proper sanitization, a user could attempt to inject malicious parameters into the data-fetching tools.
- Ingestion points: User-provided 'Holding list' and 'Cash position' collected in
SKILL.md(Step 1). - Boundary markers: None provided in the prompt logic or script execution instructions.
- Capability inventory: Subprocess execution of
views_runner.pywith user-supplied arguments as seen inreferences/data-queries.md. - Sanitization: No validation, escaping, or filtering of the user-provided ticker symbols or quantities is specified before command execution.
Audit Metadata