The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill exposes an obsidian eval command which allows for the execution of arbitrary JavaScript code strings within the context of the running Obsidian application. This can be used to access the application's internal API, modify data, or potentially perform unauthorized actions if the input is not strictly controlled.
[INDIRECT_PROMPT_INJECTION]: The skill retrieves data from external sources (Obsidian notes) using commands like obsidian read and obsidian search. There is a risk that malicious instructions contained within these notes could influence the agent's subsequent actions, especially given the skill's high-privilege capabilities like code execution and file modification.
Ingestion points: Data is ingested through obsidian read, obsidian search, and obsidian dev:console (SKILL.md).
Boundary markers: There are no instructions provided to the agent to treat vault content as untrusted data or to ignore instructions found within files.
Capability inventory: The skill allows for file creation, modification (append, property:set), and arbitrary JavaScript execution (eval).
Sanitization: No evidence of sanitization or escaping of note content before it is processed or used in subsequent commands.
[COMMAND_EXECUTION]: The skill operates by executing commands via the obsidian CLI. This requires the environment to have the CLI tool pre-installed and properly configured.

obsidian-cli