A secure, panel-hosted AI assistant for managing Pterodactyl servers with approvals.

Server AI Assistant adds a secure, permission-scoped AI agent directly to the Pterodactyl Panel. It gives customers a conversational way to manage their game servers without leaving the panel, while keeping every action restricted to their authenticated Pterodactyl account, assigned permissions, and selected server. Key features include: • Manage server files using the Wings file API • Read, create, edit, move, copy, rename, download, and delete files • Search files through supported native endpoints or controlled recursive discovery • View console output and send console commands • Start, stop, restart, and manage server power • Create, inspect, and manage backups • View and manage databases • Inspect allocated network ports • Review and modify startup settings • Diagnose crashes, configuration problems, and performance issues • Require explicit user approval before sensitive or destructive actions • Display live responses, tool activity, reasoning status, and worker-agent activity • Preserve conversations with compressed and encrypted database storage Every conversation is isolated to the current panel user and server UUID. The assistant cannot freely access the panel host or another customer’s server. Before executing an action, the backend revalidates the authenticated user, server context, and required Pterodactyl permission. Unauthorized requests are rejected, and file operations are performed through Wings rather than by granting the AI direct filesystem access. Administrators can choose how the assistant is powered: • Individual ChatGPT account authentication • OpenAI API • Anthropic • OpenRouter • xAI • Google • Z.AI • Moonshot AI • Multiple enabled providers and models at the same time The administration interface also provides controls for: • Available providers and models • Per-user token allowances • Allowance reset periods • Maximum output tokens • Runtime and concurrency limits • Session limits and retention • Approval and safety policies • Provider credentials • Assistant name, profile picture, and branding • Runtime health and diagnostics Users receive persistent private sessions across their permitted servers, streaming responses where supported, model and reasoning controls, conversation history, remaining allowance information, context usage, and compaction status. Server AI Assistant turns routine server administration into a guided chat experience while retaining Pterodactyl’s existing permission model, Wings-based node access, explicit approvals, encrypted history, isolated runtimes, and administrator-controlled usage limits.