The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes content from an Obsidian vault that may contain malicious instructions.
Ingestion points: Vault data is ingested through obsidian_rag_query and obsidian_read_note as described in SKILL.md.
Boundary markers: There are no instructions to use delimiters or ignore instructions found within the retrieved note content.
Capability inventory: The skill uses tools for semantic search, note reading, and link traversal within the local file environment.
Sanitization: No sanitization or validation of note content is performed before the information is synthesized and presented to the user.

research