The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides an obsidian eval command that allows the execution of arbitrary JavaScript code directly within the Obsidian application context. This is a high-privilege operation that could be exploited to perform unauthorized actions within the app if the input code is derived from untrusted sources.
[COMMAND_EXECUTION]: The skill uses the obsidian CLI to perform a wide range of system and application-level tasks, including file creation, deletion, and modification. It also includes commands to capture screenshots (dev:screenshot) and inspect application DOM elements (dev:dom), which provides the agent with extensive control over the user's workspace environment.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it facilitates the ingestion of untrusted data from the user's Obsidian vault.
Ingestion points: Vault content is ingested through obsidian read, obsidian search, obsidian dev:dom, and obsidian dev:console (as seen in SKILL.md).
Boundary markers: There are no instructions or delimiters defined to help the agent distinguish between its system instructions and the content of the notes being read.
Capability inventory: The skill possesses significant write and execute capabilities, including obsidian eval, obsidian create, and obsidian append.
Sanitization: There is no evidence of sanitization or validation of the content retrieved from the vault before it is processed or used in further commands.

obsidian-cli