belt

The belt skill provides AI agents with access to the belt CLI, a cloud platform command-line interface designed specifically for AI automation. This skill solves the challenge of integrating hundreds of AI models, persistent knowledge management, and reusable workflows into agent operations through a single lightweight binary with no runtime dependencies. Agents can dynamically discover and execute AI capabilities without requiring pre-configured integrations or complex setup.

The skill's main features include running over 250 AI applications across domains like image generation (Flux Pro), video creation (Veo 2), web search (Tavily, Exa), and audio synthesis (ElevenLabs). It provides persistent agent memory through a knowledge system that supports semantic search and typed entries (concepts, observations, references, preferences). The skill registry enables discovery and execution of reusable workflows either on-demand or as installed capabilities. Additionally, it supports MCP (Model Context Protocol) server connections for extending agent capabilities with external tools and services.

This skill targets AI agents and autonomous systems that need to orchestrate multiple AI models, maintain context across sessions, and leverage community-built workflows. It's particularly valuable for agents performing creative tasks (generating images or videos), research and information gathering (web search, knowledge management), or complex multi-step operations that benefit from accessing pre-built skills. The unified command structure and JSON output support make it ideal for programmatic use in agent architectures.