The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts (tools/financial_rigor.py, tools/terminal_value.py, tools/report_audit.py) to perform precise financial valuations and verify data integrity. These commands are used for legitimate analytical purposes and rely on scripts within the project's own repository.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external financial websites (e.g., macrotrends, stockanalysis, aastocks, 东方财富) to populate its analysis.
Ingestion points: External financial data sources are accessed during the "Data Collection" step in SKILL.md.
Boundary markers: No specific prompt delimiters or warnings for the agent to ignore embedded instructions in the scraped data are present.
Capability inventory: The skill utilizes Bash for command execution, Task for sub-agent management, and Write for report generation.
Sanitization: The skill mitigates risks through a mandatory "Data Cross-Validation" step requiring two independent sources and programmatic verification via specialized tools to ensure data points are numeric and consistent.
[DATA_EXPOSURE]: The skill creates a detailed investment report at ~/[公司名]投资研究报告.md. This is the intended primary output of the skill and does not involve the exfiltration of sensitive system or user data.

investment-research