openai-knowledge

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent to use the OpenAI Developer Documentation MCP tools (e.g., mcp__openaiDeveloperDocs__fetch_openai_doc) to fetch markdown at runtime from the configured MCP server (https://developers.openai.com/mcp), and that fetched content is injected into the agent context to control prompts—constituting a required runtime external dependency.