The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute emacsclient --eval. This allows the agent to execute arbitrary Emacs Lisp code within the user's Emacs session.
[DYNAMIC_EXECUTION]: Instructions in SKILL.md direct the agent to construct a Lisp expression string containing a directory path and a list of filenames. This expression is then executed via emacsclient. If the inputs (directory or filenames) contain Lisp delimiters or escape characters, it could lead to the execution of unintended Lisp functions.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect injection through its processing of data from the conversation context.
Ingestion points: The skill reads file names and directory paths from the "most recent interaction" (generated files, tool outputs, etc.) in SKILL.md.
Boundary markers: There are no explicit instructions or delimiters provided to the agent to escape or sanitize the filenames before interpolating them into the Lisp (quote ...) block.
Capability inventory: The skill uses the Bash tool to run shell commands in SKILL.md and defines Lisp logic in agent-skill-dired.el to manipulate the Emacs file manager (Dired).
Sanitization: No sanitization or validation of the input strings is performed before they are embedded into the shell command.

dired