ag2-network-quickstart
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates multi-agent interactions where agents process messages sent by others, creating an inherent surface for indirect prompt injection. \n
- Ingestion points: Agent notify handlers receive text content from
EV_TEXTenvelopes as described inSKILL.md.\n - Boundary markers: The provided code examples do not explicitly demonstrate the use of message delimiters to distinguish between system instructions and untrusted data content from other agents.\n
- Capability inventory: Registered agents can utilize tools such as
sayorend_conversationand access knowledge stores likeMemoryKnowledgeStore(SKILL.md).\n - Sanitization: No specific content validation or sanitization logic is shown in the basic quickstart examples.
Audit Metadata