The Agent Skills Directory

[SAFE]: The skill uses standard agent tools such as read_file, write_file, patch, and search_files to interact with a local directory. This is consistent with its stated purpose of managing an Obsidian vault.
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes untrusted data from markdown notes.
Ingestion points: Content from markdown files within the vault is ingested via read_file and search_files tools.
Boundary markers: The instructions do not define specific delimiters or guidelines to ignore instructions that might be embedded within note content.
Capability inventory: The agent has access to terminal, read_file, write_file, patch, and search_files capabilities.
Sanitization: No validation or sanitization of the note content is specified before the data enters the agent's context.

obsidian