The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting and analyzing data from external, untrusted sources such as SEC filings, company websites, and earnings transcripts.
Ingestion points: Data is collected in references/task1-company-research.md (Step 1: Initial Data Collection) and references/task2-financial-modeling.md (Step 1: Extract Historical Financials).
Boundary markers: The instructions do not define specific boundary markers or system prompts to ignore potential instructions embedded in the external research data.
Capability inventory: The agent utilizes the DOCX skill, XLSX skill, and Python execution environments (matplotlib/seaborn) to process this data. It also uses the Read tool to access markdown files.
Sanitization: There is no evidence of explicit sanitization or filtering of the external content before it is processed into the final report.
[EXTERNAL_DOWNLOADS]: The chart generation workflow in references/task4-chart-generation.md requires the installation of well-known Python packages including matplotlib, seaborn, pandas, numpy, and plotly. These are standard tools for financial data visualization and are retrieved from official package registries.
[DYNAMIC_EXECUTION]: The skill generates and executes Python code to produce complex financial visualizations such as DCF sensitivity heatmaps and valuation football field charts. The code is constructed using static templates that are populated with data extracted from the financial model.

initiating-coverage