multi-agent-chat
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a protocol for agents to interact in shared communication environments (Discord, Slack, Teams), which constitutes an attack surface for indirect prompt injection from untrusted participants.
- Ingestion points: Incoming messages from multi-user chat channels referenced in SKILL.md.
- Boundary markers: Not present; the instructions do not suggest using delimiters to isolate external message content.
- Capability inventory: None; the skill does not define specific tools, scripts, or subprocess calls, relying instead on the agent's native platform interface.
- Sanitization: Not present; the skill lacks instructions for validating or escaping content from human or other agent participants.
- [SAFE]: The content is purely instructional and aimed at improving agent coordination and token efficiency. No malicious patterns, obfuscation, or dangerous execution vectors were found.
Audit Metadata