The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on executing local Ruby scripts (e.g., lib/buttercut/library.rb) via the shell to manage library state, metadata, and backups. These commands incorporate user-supplied parameters such as library names and file paths.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources that could contain embedded instructions intended to influence the agent's behavior.
Ingestion points: User-provided file paths and library names, as well as metadata stored in library.yaml files.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore potential commands embedded in filenames or metadata.
Capability inventory: The skill can execute Ruby scripts, write to metadata files, and generate library backups.
Sanitization: There is no mention of sanitizing file paths or metadata content before it is processed by the Ruby scripts or downstream sub-agents.

process-library