The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill includes an eval command (obsidian eval code="...") that allows the agent to execute arbitrary JavaScript within the context of the running Obsidian application. While intended for development and debugging, this provides a mechanism to run unvalidated code that can access internal application data or modify application behavior.
[COMMAND_EXECUTION]: The skill acts as a wrapper for the obsidian command-line interface. It executes shell commands to perform vault operations, file management, and developer tasks, granting the agent significant control over the local Obsidian environment.
[INDIRECT_PROMPT_INJECTION]: The skill features commands to read and search vault content (read, search, backlinks), which creates an ingestion surface for untrusted data. If a note contains malicious instructions, it could influence the agent's behavior during processing.
Ingestion points: Commands such as obsidian read, obsidian search, and obsidian daily:read in SKILL.md bring external file content into the agent's context.
Boundary markers: The skill instructions do not specify any delimiters or ignore-instructions for the data read from the vault.
Capability inventory: The skill can execute shell commands via the obsidian CLI, write to the file system (create, append, dev:screenshot), and execute arbitrary JavaScript (eval).
Sanitization: No sanitization or validation of the content retrieved from the vault is mentioned.

obsidian-cli