The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes an external obsidian CLI tool to interact with a running Obsidian instance, enabling vault management and plugin development workflows.
[DYNAMIC_EXECUTION]: Includes an obsidian eval command that allows the execution of arbitrary JavaScript within the application context. This feature, while intended for debugging and plugin development, allows for runtime code execution.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection. 1. Ingestion points: Reads vault content via obsidian read and obsidian search (SKILL.md). 2. Boundary markers: None present to distinguish data from instructions. 3. Capability inventory: Can execute code (obsidian eval), write files (obsidian create), and capture visuals (obsidian dev:screenshot). 4. Sanitization: No sanitization or validation of vault content is mentioned.
[DATA_EXFILTRATION]: Provides commands to read note contents (obsidian read) and capture visual data (obsidian dev:screenshot), exposing potentially sensitive information stored within Obsidian vaults to the agent's context.

obsidian-cli