MCP-Server

CRITICAL VULNERABILITY: The `mcp-client/ai.go` explicitly disables SSL/TLS certificate verification by setting `TLSClientConfig: &tls.Config{InsecureSkipVerify: true}` when connecting to the Google Gemini API. This makes the connection highly susceptible to Man-in-the-Middle (MITM) attacks, allowing an attacker to intercept, read, and modify communication (including the GEMINI_API_KEY) without detection. This is a severe security flaw. Additionally, the `mcp-server/k8s.go` executes `kubectl` commands based on input from the client (and implicitly, the AI). While `exec.CommandContext` mitigates simple shell injection, the server acts as a trusted executor for arbitrary YAML content (from files or AI-generated manifests). There is no inherent validation of the YAML content itself by the server before applying it, making the system vulnerable to malicious policy definitions or resource deployments if the AI hallucinates harmful content or the client is compromised.