The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted EPUB files to generate markdown intended for LLM context, creating a surface for indirect prompt injection.
Ingestion points: User-provided EPUB file path ($1) referenced in SKILL.md.
Boundary markers: The skill does not implement delimiters or instructions to the LLM to ignore potentially malicious content within the converted markdown.
Capability inventory: The skill utilizes Bash, Read, and Write tools as specified in the SKILL.md frontmatter.
Sanitization: No sanitization or content filtering is performed on the data extracted from the EPUB files.
[COMMAND_EXECUTION]: The skill executes a shell command to perform the file conversion.
Evidence: Runs epub-to-markdown convert "$1" with the --multiple-files flag as described in SKILL.md.
Note: The input variable is quoted, and the dependency is managed through a vendor-specific installation skill.

convert-epub