multiplayer-sync

Pass

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill describes using fetch and WebSocket for network communication, which is necessary for multiplayer functionality. The examples provided use well-known placeholder domains (e.g., example.com) and link to official Decentraland test scenes on GitHub, which are trusted sources according to the author context.
  • [DATA_EXFILTRATION]: The instructions cover signedFetch, a standard Decentraland API used to send authenticated requests containing the user's wallet address to an external server. This is a legitimate architectural pattern for verifying player identity in leaderboards and reward systems.
  • [INDIRECT_PROMPT_INJECTION]: The skill highlights multiple ingestion points for external data from peers or servers, creating an attack surface for indirect prompt injection if the data is not handled carefully.
  • Ingestion points: Data received via MessageBus.on in SKILL.md, WebSocket.onmessage in networking-patterns.md, and fetch responses in networking-patterns.md.
  • Boundary markers: The examples do not specify boundary markers or instructions to ignore embedded commands within the network payloads.
  • Capability inventory: The skill facilitates network communication and engine state updates across all scripts. It does not perform high-risk operations like file system access or subprocess execution.
  • Sanitization: While the examples demonstrate JSON parsing and type handling, they do not show explicit sanitization or filtering of the content received from external sources.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 22, 2026, 11:07 PM