The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses bash commands (ln -sf, cp) to manage configuration files. There is a potential risk of command injection if the <name> parameter provided by the user is not properly sanitized before being interpolated into the shell command (e.g., ln -sf ~/.obsidian-wiki/config.<name> ...).
[DATA_EXPOSURE]: The skill reads and displays the contents of configuration files located at ~/.obsidian-wiki/config.*. While it includes logic to redact lines containing API_KEY or SECRET, it still exposes the rest of the configuration data to the agent's context and the user session.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from local configuration files which could be modified by external processes.
Ingestion points: ~/.obsidian-wiki/config.* (read during list, show, and switch operations).
Boundary markers: None. The instructions do not specify any delimiters or warnings to ignore instructions embedded within the configuration files.
Capability inventory: File read (implicitly via description), file write (via cp and value updates), and command execution (ln, cp).
Sanitization: Limited to redacting specific secret keywords (API_KEY, SECRET) during the "Show" command output.

wiki-switch