neon-ai-gateway

Pass

Audited by Gen Agent Trust Hub on Sep 23, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a technical manual for the Neon AI Gateway. It follows best practices for secret management by instructing users to use environment variables (NEON_AI_GATEWAY_TOKEN) and standard .env file integration via the Neon CLI, rather than hardcoding sensitive information.
  • [EXTERNAL_DOWNLOADS]: Fetches documentation, configuration files, and model catalog data from official sources including neon.com, models.dev, and github.com/neondatabase. These targets are well-known services belonging to the organization providing the skill.
  • [COMMAND_EXECUTION]: Provides examples for using the neon CLI tool (e.g., neon skills -s neon -y, neon deploy, neon env pull). These are standard administrative operations for users of the platform and do not represent malicious behavior.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of agents that process user input via external LLMs and tool-calling interfaces (e.g., in the AI SDK and Mastra examples). While this architecture creates a surface for indirect prompt injection, the skill itself is purely instructional and does not contain exploitable logic or unsafe data handling.
  • Ingestion points: Untrusted user input enters the agent context through the 'messages' and 'prompt' parameters in code examples.
  • Boundary markers: Not explicitly defined in the minimal code snippets; users are expected to implement these within their application logic.
  • Capability inventory: The skill enables network operations (calling the LLM gateway) and local execution via tool-calling functions (e.g., the 'listTodos' tool calling a database).
  • Sanitization: Not present in the high-level code examples provided.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 23, 2026, 04:55 AM
Security Audit — agent-trust-hub — neon-ai-gateway