The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/test_saas_metrics.py file uses subprocess.run to execute the local saas-metrics.py script for testing purposes. The command execution is confined to the skill's local directory and uses the authenticated Python environment.
[DYNAMIC_EXECUTION]: In scripts/test_saas_metrics.py, the importlib library is used to dynamically load the saas-metrics.py script. This is a standard approach to import modules with filenames containing hyphens during testing.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process financial metrics provided by users or documents, which constitutes a potential attack surface. Ingestion points: Numerical data fields for MRR, churn, and growth metrics processed by scripts/saas-metrics.py. Boundary markers: The instructions in SKILL.md advise the agent to separate assumptions from outputs, which helps maintain context. Capability inventory: The skill performs local script execution and module loading within its test suite. Sanitization: The calculation script uses argparse with strict float typing for all numerical inputs, which prevents simple text-based prompt injection through metric parameters.

financial-modeling