The Agent Skills Directory

EXTERNAL_DOWNLOADS (LOW): The script scripts/fmp_client.py performs network requests to https://financialmodelingprep.com to retrieve earnings calendars and historical price data. While this domain is not on the predefined whitelist, these operations are central to the skill's purpose and are performed using standard HTTP libraries.
DATA_EXPOSURE (SAFE): The skill accesses a sensitive API key through the FMP_API_KEY environment variable. This is consistent with security best practices for credential management in agent skills.
INDIRECT PROMPT INJECTION (SAFE): The skill ingests external data from the FMP API and local JSON files.
Ingestion points: fmp_client.py (API data) and screen_pead.py (Mode B JSON input).
Boundary markers: Not explicitly observed in logic, but data is treated as numeric/structured content.
Capability inventory: The skill performs calculations and generates reports; it does not contain code execution capabilities like eval or subprocess that could be triggered by data.
Sanitization: Values are converted to floats and rounded during scoring calculations.

pead-screener