The Agent Skills Directory

[SAFE]: The skill performs legitimate financial analysis by monitoring and quantifying silver stock levels. Its operations are transparently documented in the methodology and workflow files.
[EXTERNAL_DOWNLOADS]: The skill utilizes the webdriver-manager library to automatically download and manage browser drivers (e.g., ChromeDriver). This is a standard and safe practice for automation tasks involving Selenium.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests data from external financial websites (CEIC Data, SGE, SHFE). However, the evidence shows that this data is processed strictly as numerical input for quantitative analysis (Z-score calculations, drain rate metrics), which minimizes the risk of executing embedded instructions.
Ingestion points: External URLs processed in scripts/drain_detector.py and various fetch_*.py scripts.
Boundary markers: Not explicitly defined for the scraped data.
Capability inventory: File writing (CSV results) and network access for data fetching.
Sanitization: The scripts use standard parsing (e.g., pd.read_csv, re.findall) to extract specific numerical values.
[COMMAND_EXECUTION]: The skill executes standard Python scripts and library calls for its primary function. No evidence of arbitrary command execution or privilege escalation was found.

detect-shanghai-silver-stock-drain