The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted text provided by the user or read from files, creating an attack surface for indirect prompt injection.
Ingestion points: Content is ingested from user prompts and through the view_file tool when operating in File or Repo Audit modes.
Boundary markers: The instructions do not implement strict delimiters or explicit warnings for the agent to ignore instructions embedded within the target prose.
Capability inventory: The skill has access to tools that can modify the file system, including write_to_file and replace_file_content.
Sanitization: There are no documented mechanisms to sanitize the ingested text for malicious instructions before processing.
[EXTERNAL_DOWNLOADS]: The documentation references installation via npx skills add harshsinghmp/muse-skills, which retrieves the skill or the 'Muse Skills' suite from the author's GitHub repository.

humanize