The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions in SKILL.md and references/docstrings.md involve reading and normalizing Objective-C source code. This ingestion of external data constitutes an attack surface for indirect prompt injection, as there are no explicit boundary markers or sanitization steps to distinguish code from potentially malicious instructions embedded in comments. (Evidence: Ingestion points in SKILL.md workflow and references/docstrings.md targets; Boundary markers: Absent; Capability: Source code modification; Sanitization: Absent).
[EXTERNAL_DOWNLOADS]: The references/docstrings.md file links to official Apple documentation and the realm/jazzy GitHub repository. These are well-known and trusted developer resources, and their inclusion is standard for a skill focused on Objective-C documentation.

objective-c-docstrings