The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes the Bash tool to run Python scripts using shell interpolation of user-supplied arguments such as vault paths and filenames. If these inputs contain unescaped shell metacharacters or quotes, it could lead to arbitrary command execution on the host system.
[REMOTE_CODE_EXECUTION]: The skill uses a heredoc to write a complete Python program to the /tmp directory at runtime and then executes it. This dynamic code generation and execution pattern increases the attack surface as the code is not statically defined and depends on successful shell execution of generated content.
[PROMPT_INJECTION]: The skill reads and parses local Markdown files, extracting text to identify image links and printing the results back to the agent's context. This ingestion of untrusted data without sanitization enables indirect prompt injection, where a malicious note could attempt to manipulate the agent's behavior by embedding instructions in the script's output reported back to the LLM.

export-vault-note