pubnub-app-context

Pass

Audited by Gen Agent Trust Hub on Jun 30, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill correctly defines implementation patterns for the PubNub Objects API. It uses legitimate vendor-specific MCP tools and documentation references.
  • [DATA_EXPOSURE]: No hardcoded credentials or sensitive configuration files were detected. The skill uses placeholders for API keys and tokens in its examples.
  • [PROMPT_INJECTION]: The skill manages external data such as user profiles and channel metadata, which constitutes a standard indirect prompt injection surface common to data-management applications. No evidence of malicious exploitation or bypass instructions was found.
  • Ingestion points: Metadata fields in setUUIDMetadata, setChannelMetadata, and setMemberships (SKILL.md, references/users.md).
  • Boundary markers: Not explicitly implemented in the provided prompt logic.
  • Capability inventory: Uses the manage_app_context tool for standard CRUD operations on PubNub metadata.
  • Sanitization: The skill does not prescribe specific input sanitization for metadata fields, which is consistent with its role as an API implementation guide.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 30, 2026, 12:42 PM
Security Audit — agent-trust-hub — pubnub-app-context