The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates reading from and writing to a file system vault, creating a potential surface for indirect prompt injection if note content contains malicious instructions.
Ingestion points: The skill uses search_notes and get_tree within the SKILL.md instructions to access the personal/ directory.
Boundary markers: No explicit delimiters or instructions to ignore embedded content are defined within this adapter, though it references the hatchdoor skill for frontmatter management.
Capability inventory: The skill utilizes create_note, edit_note, replace_section, and append_to_note to modify vault data.
Sanitization: There is no evidence of content sanitization or validation before data is processed or written back to the vault.

effort-store-vault