Buoyant adds MCP support for Linkerd
Buoyant is the creator of Linkerd, an open source service mesh for cloud-native applications.
Linkerd provides lightweight tools for securing, observing and managing “service-to-service communication” in Kubernetes environments.
What does that mean? In basic terms, service-to-service communication here means the ability for separate components of a microservices application to securely communicate.
Now looking to enhance functionality and increase system strength, Buoyanthas announced upcoming support for Model Context Protocol (MCP) in Linkerd to extend its core service mesh capabilities to this new type of agentic AI traffic.
This capability is designed to offer an “observable foundation” for MCP traffic in Kubernetes environments
What is MCP?
MCP is a protocol that connects AI models with external data sources and tools in a standardised way.
As defined by its creators, Anthropic, Model Context Protocol is an open standard that enables developers to build secure, two-way connections between their data sources and AI-powered tools. The architecture is straightforward: developers can either expose their data through MCP servers or build AI applications (MCP clients) that connect to these servers.
Why is this so important?
Because agentic communication is unique compared to the API traffic that dominates enterprise environments today, the communication pulls from past interactions, takes place over a persistent session and may require complex, multi-task steps.
Compared to standard enterprise workloads, AI workloads are unpredictable because they:
- Introduce novel security challenges
- Can create massive spikes in traffic
- Lead to a high-risk management environment.
In response, Buoyant says that most organisations are not equipped to manage. The rate at which AI initiatives are being pushed at an enterprise level amplifies the urgent need for more security and observability for application traffic.
“Enterprises are eager to innovate with AI, but they can’t do so at the expense of their security posture and application reliability,” said William Morgan, CEO of Buoyant. “Linkerd solves this problem by extending its proven capabilities to MCP traffic. We’re not just enabling AI adoption, we’re giving organizations the tools to accelerate their usage with confidence.”
Buoyant claims managing complex network traffic is Linkerd’s “proven area of expertise” and with the introduction of MCP support, Linkerd will provide the same visibility, access control and traffic shaping capabilities that users rely on today for their existing set of protocols, extended to agentic traffic.
Terrific on traffic
Specifically, this upcoming offering will provide:
Observability of MCP traffic with metrics on resource, tool and prompt usage, including failure rates, latencies and volume of data transmitted. This allows enterprises to build an operational view of agentic behaviour while monitoring and alerting on unusual activity.
Security for MCP traffic through fine-grained authorisation policies for all MCP calls, using Linkerd’s existing zero-trust framework built on cryptographic workload identity. This allows enterprises to restrict access to specific tools or resources exposed by MCP servers based on the identity of the agent and implement a zero-trust approach to agentic security.
“The security concerns around MCP were initially a big factor as we started considering rolling out AI more broadly across the organisation,” said Blake Romano, senior engineer at Imagine Learning. “Because we already use and trust Linkerd’s enterprise capabilities, especially its strong security posture and built-in observability, it removed a major barrier to adoption. Having clear visibility into our MCP communications gives us the confidence to innovate faster, safer and more reliably.”
By integrating MCP support at the service mesh layer, Linkerd says it eliminates the need for additional, specialised tooling and provides a single pane of glass for all traffic.
“Linkerd is the only service mesh that offers support for MCP integrated into the core functionality, meaning that platform teams can benefit from using the same trusted workflows and consistent policies in Linkerd to manage both traditional and AI workloads,” said Morgan.
MCP support will be fully available in 2026 in both open source Linkerd and the enterprise distribution from Buoyant.
