The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from an external file named company-context.md to provide relevant financial advice. This represents a potential attack surface for indirect prompt injection if the source file contains malicious instructions.
Ingestion points: The agent is instructed to read company-context.md before responding (SKILL.md).
Boundary markers: No specific delimiters or "ignore instructions" warnings are defined for the ingestion of this context file.
Capability inventory: The skill has capabilities to execute local Python scripts (python scripts/*.py) and write output artifacts (SKILL.md).
Sanitization: The skill employs a "Chain of Thought" reasoning technique and an "Internal Quality Loop" to self-verify output, which serves as a functional oversight mechanism.
[COMMAND_EXECUTION]: The skill provides documentation for executing local Python scripts to perform financial calculations. These executions are limited to the skill's own script directory and do not involve remote code downloads or elevated privileges.
Evidence: The "Quick Start" section demonstrates using python scripts/burn_rate_calculator.py, python scripts/unit_economics_analyzer.py, and python scripts/fundraising_model.py (SKILL.md).

cfo-advisor