The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill exposes the obsidian eval code="..." command, which allows for the execution of arbitrary JavaScript within the Obsidian application context. This functionality is intended for plugin development and testing but represents a powerful execution surface if the agent interpolates untrusted data into the code string.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates reading content from user vaults through commands such as obsidian read and notesmd-cli print. This creates an attack surface where instructions embedded in note files could potentially manipulate the agent's actions.
Ingestion points: Vault files accessed via obsidian read, notesmd-cli print, obsidian daily:read, and obsidian search commands.
Boundary markers: None; the skill instructions do not prescribe the use of delimiters or specific warnings to ignore instructions within the retrieved content.
Capability inventory: The skill provides high-privilege capabilities including arbitrary command execution, dynamic JavaScript evaluation (obsidian eval), and comprehensive file system operations (create, append, move, delete).
Sanitization: There is no mention of sanitizing or validating note content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill's primary function is the execution of terminal-based commands (obsidian and notesmd-cli) to manage the Obsidian application and its underlying data structures.

obsidian-clis