The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use shell commands (test, mv) for legitimate purposes, such as checking for the existence of preference files (EXTEND.md) and managing backup versions of generated analysis, prompt, and image files.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided text or files for analysis and transformation, which constitutes a standard attack surface for indirect prompt injection.
Ingestion points: Arbitrary text provided via copy-paste or local file paths (e.g., source.md).
Boundary markers: The instruction set uses a template (references/base-prompt.md) with a {{CONTENT}} placeholder, though it lacks specialized security delimiters.
Capability inventory: Shell command execution (test, mv), local file read/write operations, and invocation of external image generation skills.
Sanitization: No specific input sanitization or filtering logic is documented for the source content before it is interpolated into the final prompt.
[SAFE]: File system access is restricted to the skill's own output directory (infographic/{topic-slug}/) and a dedicated configuration path ($HOME/.baoyu-skills/), representing best practices for application-specific data storage and preference management.

baoyu-infographic