scene-runtime
Pass
Audited by Gen Agent Trust Hub on Feb 25, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by processing external data and possessing high-privilege capabilities.
- Ingestion points: External data enters the system context via
fetch,WebSocket, andreadFileoperations described inSKILL.md. - Boundary markers: The documentation does not specify or demonstrate the use of delimiters or instructions to distinguish untrusted external content from system prompts.
- Capability inventory: The runtime provides functions for network communication (
fetch,signedFetch), player manipulation (movePlayerTo,teleportTo), and browser interaction (openExternalUrl). - Sanitization: The provided snippets show data being processed via
JSON.parseandTextDecoderwithout explicit validation or sanitization steps before the data is used.
Audit Metadata