The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses obsidian-cli or obsidian to manage the vault. It checks for their existence using command -v and falls back to filesystem tools like rg if needed. These are standard local operations for the skill's purpose.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill interacts with files within a directory specified by the OBSIDIAN_KNOWLEDGE_VAULT environment variable. It does not access sensitive system paths and lacks network capabilities for exfiltration.
[INDIRECT_PROMPT_INJECTION]: The skill processes content from an Obsidian vault, which could contain untrusted data from external sources (Literature Notes). While boundary markers for this content are absent, this is inherent to the nature of a note-taking tool.
Ingestion points: Reads .md files from the local vault via the read command or direct filesystem access.
Boundary markers: None explicitly provided for interpolated note content.
Capability inventory: Filesystem read/write operations and execution of the obsidian-cli tool.
Sanitization: None mentioned for processed note content.

knowledge