a2ui-project/a2ui
Open standard and renderer libraries letting AI agents describe UI as safe, declarative JSON that client apps render natively -- on Flutter, Angular, Lit, SwiftUI and more.
Awesome AI Agents › Communication Protocols
Novu provides a comprehensive open-source communication infrastructure designed to streamline notification delivery for products and AI agents. It centralizes messaging across diverse channels like Inbox, Email, SMS, Push, Chat, Slack, Microsoft Teams, and Telegram, utilizing a single API and a unified conversation model. For products, Novu offers an integrated platform to send multi-channel notifications, manage workflows, create digests, and embed a real-time Inbox component, abstracting away the complexity of integrating multiple providers. For AI Agent Communication Infrastructure (ACI), Novu enables agents to interact with users on various communication channels by providing a unified model for bidirectional messaging. It handles inbound messages, normalizes them, routes them to the agent, and dispatches agent replies, eliminating the need for per-platform webhook maintenance. Novu includes features like an embeddable Inbox for web apps, a no-code email editor, and user preference management. It aims to transform static notifications into meaningful dialogues, catering to the rising user expectation for conversational interactions.
https://github.com/novuhq/novu
Open standard and renderer libraries letting AI agents describe UI as safe, declarative JSON that client apps render natively -- on Flutter, Angular, Lit, SwiftUI and more.
Agent Payments Protocol (AP2) is a project focused on creating a secure and interoperable protocol for AI-driven payments, providing code samples and demos to facilitate development and integration.
Official sample repository for the Agent2Agent (A2A) protocol, with runnable agent and host implementations in Python, Go, Java, JavaScript and C# built on the official A2A SDKs.
AgentNetworkProtocol (ANP) is an open-source protocol designed to enable secure, decentralized, and efficient communication and collaboration among intelligent agents, aiming to become the HTTP of the Agentic Web era.
Research implementation of Cache-to-Cache, a method that lets two large language models exchange semantics directly by projecting and fusing their KV-caches instead of passing generated text between them.
Python reference implementation of the Universal Tool Calling Protocol, a plugin-based standard that lets agents discover and call existing tools directly over HTTP, SSE, CLI and other transports.
Specification and reference libraries for the x402 extension to the Agent-to-Agent protocol, letting one agent charge another for a service through on-chain stablecoin payments in a three-message flow.
a2a-java is the official Java SDK for the Agent2Agent protocol, providing client and server modules that expose agentic Java applications over JSON-RPC, gRPC and REST transports.