The Agent Skills Directory

[SAFE]: The skill consists entirely of natural language instructions for the agent's behavior. No scripts, executables, or external dependencies are included.
[DATA_EXPOSURE]: There are no commands that access sensitive files or perform network operations. The skill is limited to processing text provided by the user within the chat context.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data (user-provided drafts). While it lacks explicit boundary markers (e.g., delimiters) to separate instructions from user content, the agent has no dangerous capabilities (like network access or file system writes) to exploit. The ingestion occurs through natural language processing of user inputs.

ask-hr