The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the bash tool to check for the existence of an EXTEND.md preference file in the project directory or the user's home folder. This is a standard and safe method for loading user-defined configurations.
[PROMPT_INJECTION]: While the skill ingests untrusted user data, it employs a rigorous transformation process into a structured markdown format before prompt generation. This instructional design approach acts as a buffer against indirect prompt injection.
[SAFE]: All layout and style definitions are stored locally within the references/ directory. The skill does not perform arbitrary remote downloads or execute unverified external scripts.
[SAFE]: There are no hardcoded credentials, API keys, or sensitive file path accesses detected in the skill's logic or metadata.

baoyu-infographic