websocket-implementation
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a communication architecture that ingests and broadcasts arbitrary user-provided text ('content') through WebSocket events. If an AI agent later consumes this message history, it could potentially be influenced by instructions embedded within the message data.
- Ingestion points: Found in the message event listeners in
SKILL.md(Node.js) andreferences/python-websocket.md(Python). - Boundary markers: None present in the provided templates; developers would need to implement their own delimiters.
- Capability inventory: The skill's capabilities are limited to real-time message broadcasting and connection management. No high-risk tools (e.g., shell execution or local file writing) are triggered by the ingested message content.
- Sanitization: The templates do not include content sanitization or validation, as the primary purpose is providing a generic transport layer.
Audit Metadata