An Extensible,
Plugin-Based AI Agent Framework
A ready-to-use, open framework featuring multimodal capabilities, swappable memory infrastructure, and a plugin ecosystem.
An Extensible,
Declarative Multimodal Capabilities
MAIAR agents natively support multimodal input and output—text, audio, vision, and beyond. The framework abstracts modality handling to a runtime level capability registry, enabling forward-compatible support for the accelerating scope of multimodal model capabilities without patching the core.
defineCapability({
id: "multi-modal-text-generation",
description: "create text using text and images",
input: z.object({
prompt: z.string(),
images: z.array(z.string()).optional()
}),
output: z.string()
});Platform Integrations
Ready to Build with MAIAR?
Dive into the docs, join our community spaces, and help shape the next wave of composable AI agents.