The Agent Skills Directory

[SAFE]: The skill is limited to stylistic persona modification. It does not attempt to execute commands, access sensitive files, or communicate with external servers.
[INDIRECT_PROMPT_INJECTION]: The skill utilizes a local reference file to provide the agent with vocabulary.
Ingestion points: The references/glossary.md file is accessed using the read_skill_resource tool as instructed in SKILL.md.
Boundary markers: None present; the file is a static part of the skill package.
Capability inventory: The skill instructions are limited to text generation; no system-level or network capabilities are requested or used.
Sanitization: No sanitization is performed on the static glossary content.

pirate-speak