A list of emerging open standards for AI agents.
Agentic AI Foundation (AAIF) Standards
The Agentic AI Foundation is a Linux Foundation project with founding contributions from Anthropic, Block, and OpenAI.
Model Context Protocol (MCP) – Anthropic
Open standard API for connecting AI agents to data sources and tools. Acts as a “USB-C port for AI” to standardize context delivery.
AGENTS.md – OpenAI
Markdown-based convention for providing project-specific instructions to coding agents. Used by 60k+ open-source projects.
Goose – Block
Open-source agent framework connecting LLMs to real-world actions via MCP. Reference implementation for building custom agents.
Other Open Standards
Agent2Agent (A2A) – Google Cloud
Protocol for peer-to-peer communication between AI agents, defining message formats and negotiation patterns.
AGNTCY – Cisco/Linux Foundation
Infrastructure for multi-agent collaboration: discovery, identity, secure messaging (SLIM protocol), and observability.
Emerging
skills.md – Currently redirects to agentskills.io
Likely to become a standard for defining agent capabilities and skills.
Reference
Based on Agentic AI Foundation: Guide to Open Standards for AI Agents (Intuition Labs, December 2025)