The Agent Skills Directory

[COMMAND_EXECUTION]: The skill interpolates user-controlled input ($ARGUMENTS) directly into shell commands within SKILL.md. This pattern is vulnerable to command injection if shell metacharacters (e.g., ;, &&, |) are included in the arguments passed to the skill.
[EXTERNAL_DOWNLOADS]: The skill includes a fallback routine to install the pyyaml library from the Python Package Index (PyPI) at runtime using pip install. While pyyaml is a well-known package, installing dependencies from public registries introduces a dependency on external infrastructure and package integrity.
[PROMPT_INJECTION]: The skill analyzes the content of markdown files in a user-specified directory, presenting a surface for indirect prompt injection where malicious files could attempt to influence the agent's behavior.
Ingestion points: The scripts/okf_validate.py script reads and parses the content of all .md files within the provided bundle directory.
Boundary markers: The skill instructions lack explicit delimiters or warnings to the agent to ignore instructions that may be embedded within the files being validated.
Capability inventory: The skill is configured to use the Bash tool for command execution.
Sanitization: The validation script uses yaml.safe_load() for YAML parsing, but it returns file contents and metadata values (e.g., error messages containing file data) to the agent without sanitization or escaping.

validate