pubnub-multiplayer-gaming

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [Data Exposure] (SAFE): The code uses placeholder strings for API keys ('pub-c-...' and 'sub-c-...') in Javascript examples, which is standard practice for documentation and does not expose real secrets.
  • [Indirect Prompt Injection] (LOW): The skill is designed to process external messages from PubNub channels, creating an attack surface for indirect prompt injection. Findings: (1) Ingestion points: pubnub.addListener message event in SKILL.md. (2) Boundary markers: Absent. The skill does not implement delimiters or 'ignore' instructions for channel messages. (3) Capability inventory: The skill allows updating internal game state via applyDelta and UI rendering via renderGame. (4) Sanitization: Absent. Data received from the network is applied directly to the local state object.
  • [External Downloads] (SAFE): The skill depends on the legitimate pubnub package, which is necessary for its functionality and sourced from standard registries.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 08:56 PM