The Agent Skills Directory

[SAFE]: The skill incorporates strong security principles, specifically a "No surprises" hard rule that forbids the inclusion of secrets, hidden network calls, or destructive defaults. It mandates explicit confirmation for any operations with external side effects.
[INDIRECT_PROMPT_INJECTION]: The skill processes user requirements and external project data during its workflow, which is a common surface for indirect prompt injection. This risk is mitigated by structured intent analysis using a "Requirements Card", multi-phase validation gates, and a requirement for a human-in-the-loop trial run before final delivery.

create-skill