The Agent Skills Directory

[SAFE]: The skill instructions define a legitimate workflow for capturing and organizing writing fragments into a single document.
[COMMAND_EXECUTION]: The skill performs file system operations (reading and writing to a Markdown file), but these are restricted to paths provided or approved by the user and do not involve executing arbitrary shell commands.
[INDIRECT_PROMPT_INJECTION]: The skill is instructed to read the current state of the file from disk before appending new content to preserve user edits. While reading external files is a standard injection surface, the skill's scope is limited to text capture, which mitigates the risk of the agent executing malicious instructions embedded in the file text.

writing-fragments