Kaltura open sources machine-readable AI agent skills, enabling developers and AI agents to rapidly build production-grade applications while accelerating the shift to agent-first software ecosystems.
Kaltura has open-sourced a suite of AI agent skills—structured, production-tested knowledge modules designed for AI coding agents such as Claude Code, OpenAI Codex, GitHub Copilot, and Cursor.
Unlike traditional documentation, these skills are machine-readable and continuously validated against production APIs. They include tested curl examples, parameter tables, error-handling logic, and security and compliance patterns, enabling AI agents to autonomously build production-grade applications.
Once deployed, the skills transform AI agents into instant platform experts, reducing development timelines from weeks to minutes and enabling end-to-end application lifecycle execution—from build and configuration to deployment and maintenance.
The capabilities span avatar-driven websites with real-time multilingual interactions, automated virtual events configured via natural language, interactive AI-powered learning environments, and fully automated content production pipelines.
Kaltura supports agent discovery and interoperability through four standards: Model Context Protocol (MCP), Agent Skills Standard, llms.txt, and Context7 integration, positioning itself within an emerging AI agent infrastructure layer. This is reinforced by its API-first architecture, comprising over 100 REST services, allowing each capability to be modularised into AI-consumable skills.
The offering is complemented by the Unisphere micro-frontend framework, which delivers over 10 embeddable widgets without build dependencies or framework lock-in.
“AI agents are rapidly becoming the primary way developers and business users interact with platforms – and most platforms aren’t ready for that shift. Kaltura’s API-first architecture, built from day one around programmable services, is perfectly suited for today’s agentic world. Each skill gets your AI agent from zero to a working, secure, best-practice digital experience in minutes,” said Zohar Babin.
“By open-sourcing these skills, we’re inviting the entire developer community to build on our platform – avatar-guided portals, intelligent content hubs, AI-powered learning environments – and making every AI agent an instant Kaltura expert.”
The skills, documentation, and developer resources are now available via open repositories, signalling a shift towards machine-native, agent-first software development.
