The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local validation script (validate-sop.sh) to verify the structure of markdown files. This script is a vendor-provided utility that uses standard command-line tools like grep for pattern matching and does not perform any network, persistence, or privileged operations.
[PROMPT_INJECTION]: The skill ingests user-provided text to generate and update .sop.md files, creating a surface for indirect prompt injection.
Ingestion points: User input is used to define SOP objectives, parameters, and step descriptions in agent-sops/*.sop.md files.
Boundary markers: The skill relies on markdown headers (e.g., ## Overview, ## Steps) to delimit sections, but does not implement specific isolation delimiters for the user-provided content.
Capability inventory: The skill utilizes file system write operations and executes a local bash-based structural validator.
Sanitization: No sanitization or filtering is performed on user inputs, as the output is treated as descriptive text for documentation purposes rather than executable code.

agent-sop-author