The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires executing local Python scripts to perform financial calculations and generate visualization charts.\n- [EXTERNAL_DOWNLOADS]: The skill fetches macro-economic and financial data from established external services. Evidence: scripts/spread_forecaster.py and scripts/plot_bloomberg_style.py download CSV data from fred.stlouisfed.org and price data via the yfinance library from Yahoo Finance.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes data retrieved from external sources, creating a potential surface for indirect prompt injection attacks if those sources were compromised. Ingestion points: Data is ingested through requests (for FRED) and yfinance (for market data) in scripts/spread_forecaster.py and scripts/plot_bloomberg_style.py. Boundary markers: The skill processes data numerically using pandas and numpy; there are no specific markers to distinguish external data within the LLM's prompt context. Capability inventory: The skill has network access for data retrieval and the ability to write analysis results and images to the local file system. Sanitization: Data is validated by coercing values to numeric types and removing missing values via standard pandas operations.

forecast-sector-relative-return-from-yield-spread