The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill's content is entirely instructional and configuration-based, focused on improving coordination and data integrity in multi-agent systems.
[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection because the validate_intent capability processes untrusted intent data provided by other agents. However, the risk is mitigated by the use of a rigid schema and a comprehensive quality checklist that prioritizes verifiable actions and evidence-based stability, making it difficult for an attacker to hijack agent behavior through intent descriptions.

intent-author