The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze and transform code provided by the user, creating a surface for indirect prompt injection. 1. Ingestion points: Code snippets provided by the user in response to trigger phrases like 'refactor this' or 'clean up this messy code' (SKILL.md). 2. Boundary markers: The skill does not provide specific instructions to the agent to ignore or delimit instructions found within the code being refactored. 3. Capability inventory: The skill instructions focus on method-level and class-level transformations across all reference files (e.g., composing-methods.md, moving-features.md), while the agent typically possesses file-system and command-execution tools to perform these refactorings. 4. Sanitization: No sanitization of user-provided code is mentioned in the instructions.

refactoring-patterns