The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest potentially untrusted data from an external file, creating a vulnerability surface where malicious instructions in that data could influence agent behavior.
Ingestion points: The skill explicitly instructs the agent to read company-context.md before producing any strategic guidance (SKILL.md).
Boundary markers: There are no specific delimiters or "ignore embedded instructions" warnings defined for the ingestion of the context file.
Capability inventory: The skill utilizes local command execution through Python scripts (tech_debt_analyzer.py and team_scaling_calculator.py) and maintains high-level decision-making authority.
Sanitization: The instructions do not specify any validation or sanitization steps for the content read from company-context.md.
[COMMAND_EXECUTION]: The skill documentation and metadata specify the execution of local Python scripts to perform core tasks.
Evidence: SKILL.md and references/advisory-workflow.md include shell commands to run scripts/tech_debt_analyzer.py and scripts/team_scaling_calculator.py. These tools are listed in the YAML frontmatter as allowed tools for the skill.

cto-advisor