The Linux Foundation recently announced it has accepted AGNTCY, an open-source project focused on cross-vendor, interoperable infrastructure for autonomous AI agents.
Cisco kicked-off the effort in March 2025 with support from LangChain and Galileo, AGNTCY has since gained backing from over 75 companies. Cisco, Dell Technologies, Google Cloud, Oracle, and Red Hat join as formative members under Linux Foundation governance.
AGNTCY delivers foundational infrastructure for the “Internet of Agents”, enabling AI agents from different frameworks, vendors, and deployment environments to discover each other, establish identity, communicate securely, and share runtime observability data.
The move to the Linux Foundation ensures vendor-neutral stewardship, supporting long-term sustainability and industry-wide adoption.
What is AGNTCY?
AGNTCY provides a vertically integrated infrastructure stack for agentic AI collaboration, comprising five core components: discovery, identity, observability, protocol integration with both MCP and A2A, and more.

Agent Discovery
AGNTCY implements Open Agent Schema Framework (OASF) and a decentralized Agent Directory. These components allow any autonomous agent to discover the capabilities, dependencies, and status of other agents. This mirrors the role of DNS in traditional networking, but with a schema-based understanding of agent function.
Agent Identity
The identity layer provides cryptographically verifiable identities and access control mechanisms for agents. These are tamper-resistant digital credentials that support cross-organizational interaction, enforce security policies, and ensure proper delegation of authority in agent workflows.
Agent Messaging
Communication between agents leverages SLIM (Secure Low-latency Interactive Messaging), a purpose-built protocol for agent communication. SLIM supports:
- Low-latency message delivery
- Human-in-the-loop interactions
- Multi-modal data exchange
- Quantum-safe encryption
This protocol provides deterministic behavior for agent interactions in time-sensitive, distributed environments.
Agent Observability
AGNTCY offers an SDK and schema to provide end-to-end observability across heterogeneous multi-agent systems. The observability layer accounts for the probabilistic and emergent behavior typical of agent-based systems, enabling real-time debugging, telemetry collection, and postmortem analysis.
Protocol Integration
AGNTCY integrates with:
- Agent2Agent (A2A): A project recently contributed to the Linux Foundation that defines agent-to-agent interaction models.
- Anthropic’s Model Context Protocol (MCP): A specification for sharing runtime context between agents and LLM-powered systems.
AGNTCY directories allow discovery of A2A agents and MCP servers, while its messaging and observability layers support these protocols natively.
Governance & Ecosystem Participation
The transition to the Linux Foundation places AGNTCY under a neutral governance model. Cisco contributed the initial codebase. Other formative members provide infrastructure, integrations, and reference implementations:
- Dell Technologies contributes interoperability frameworks across public cloud, private data center, and edge environments.
- Galileo contributes observability and evaluation tooling used in production agent systems.
- Google Cloud brings expertise in API ecosystems and platform governance.
- Oracle focuses on secure, policy-compliant agent interactions for enterprise workloads.
- Red Hat provides input on community-driven software lifecycle management and OSS compatibility.
AGNTCY has active working groups for Core, Identity, and Observability & Evaluation, which will continue development under the Linux Foundation structure.
Analysis
The AGNTCY project addresses a critical gap in agentic AI infrastructure: the lack of standardized inter-agent communication and coordination. While vendors and platforms like OpenAI, Hugging Face, and Cohere build increasingly powerful single-agent models or frameworks, cross-vendor collaboration remains fragmented. AGNTCY provides the infrastructure necessary to support agent ecosystems, not just standalone agents.
By joining the Linux Foundation, AGNTCY formalizes a neutral foundation for interoperability, potentially accelerating adoption by:
- Enterprises requiring cross-vendor agent collaboration
- Developers building complex, agentic workflows with heterogeneous components
- Vendors seeking an open framework rather than building proprietary agent orchestration systems
If AGNTCY gains broad adoption, it could become the de facto interoperability layer for agentic AI, much like Kubernetes became the standard for container orchestration. This would shift competitive differentiation toward agent capabilities and workflows rather than infrastructure, further commoditizing low-level agent management functions.
AGNTCY’s transition to the Linux Foundation formalizes an open, community-driven infrastructure layer for multi-agent AI systems. By offering integrated capabilities across discovery, identity, messaging, and observability, AGNTCY removes critical technical barriers to scaling collaborative AI workflows. Its integration with A2A and MCP standards further enhances cross-vendor compatibility.
For vendors, enterprises, and developers investing in agentic AI, AGNTCY offers a credible path to interoperability without custom integration