The Agent Skills Directory

[SAFE]: The skill implements a robust clarification loop and decision rules that prioritize safe discovery and user confirmation before performing any high-risk or irreversible actions like code commits, cloud resource creation, or data deletion.
[SAFE]: The skill uses clear boundary markers for untrusted information and encourages delegating specialized tasks (like security audits or API documentation) to specific, scoped skills, adhering to the principle of least privilege.
[SAFE]: Review of the provided evaluation cases and format specifications shows no evidence of prompt injection, credential harvesting, or remote code execution patterns. All operations are local or targeted towards well-defined project management artifacts.

spark