websockets
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and broadcasting of real-time user data, which presents a surface for indirect prompt injection if malicious instructions are embedded in messages.
- Ingestion points: WebSocket message handlers (e.g.,
chat:messageevent inSKILL.md) receive data directly from clients. - Boundary markers: The implementation examples lack explicit delimiters or instructions for the agent to ignore potentially malicious content within the message payloads.
- Capability inventory: The examples focus on network-based event emission and do not include high-risk capabilities like system command execution or unauthorized file access.
- Sanitization: While the
references/PATTERNS.mdfile recommends input validation, the core examples inSKILL.mddo not demonstrate specific sanitization or filtering of transmitted data strings.
Audit Metadata