The Agent Skills Directory

[SAFE]: The skill is a purely instructional tool for designing workflow specifications through an interview-based process.
[COMMAND_EXECUTION]: No shell commands, subprocesses, or system-level operations are defined or attempted.
[DATA_EXFILTRATION]: The skill operates entirely within the local workspace files (workflows/*.md and NOTES.md). It contains no network operations or access to sensitive global file paths.
[EXTERNAL_DOWNLOADS]: No external packages, remote scripts, or third-party dependencies are requested or executed.
[PROMPT_INJECTION]: The instructions are focused on guiding the agent's conversational style ('grilling') and do not attempt to bypass safety guidelines, override system constraints, or extract system prompts.
[DATA_EXPOSURE]: The skill manages its own documentation files and does not handle or request credentials, API keys, or private user data.

loop-me