Full-stack observability platform company Coralogix has detailed the launch of its new Model Context Protocol Server technology. The product is designed to allow third-party AI agents to connect directly to Coralogix’s observability data services to provide a deeper view into the new fabric of agentic AI connections.
Breaking down the observability viewfinder on offer here, Coralogix’s perspective purview covers logs, metrics and traces and extends to formalized security information and event management (SIEM) controls.
Newly Barrelled RUM
Additionally, the company has a wide-angle view on real user monitoring (pleasingly shortened to RUM), which we can define as performance monitoring focused on actual user clicks, inputs and other interactions (with no synthetic data incorporated, as is often the way) with an application, website or data service.
As agentic AI now becomes ubiquitously promoted and promulgated (if not actually always deployed), RUM observability enables developers and operations staff to identify bottlenecks and optimize performance. Crucially, it also offers a deeper view into how new agentic services are being used to help assess their worth.
Coralogix’s MCP Server applies to observability management across live production environments, but also extends “backwards” through application and service production, staging, prototyping, debugging and what the company loosely defines as “other environments” as well.
The new service is hoped to help organizations deploy more efficient AI agents through access to detailed observability data, which should help reduce mean time to resolution (MTTR) when errors occur. Coralogix also says it will help in terms of streamlining agent workflows and minimizing engineering overhead.
What MCP Means For DevOps
Speaking directly to DevOps.com in line with his firm’s new launch, Liran Hason, VP of AI for Coralogix, said that, “MCP fundamentally shifts DevOps by embedding AI directly into the fabric of observability, turning raw telemetry into actionable intelligence in real time. This empowers teams to move from reactive firefighting to proactive system management, automating diagnostics, speeding up incident resolution and enabling continuous reliability at scale. This interaction ultimately helps train AI agents, enhancing their capabilities and effectiveness.”
As readers will know, MCP is an open standard developed by Anthropic, the company behind Claude large language model and AI chatbot. In increasingly widespread use at the time of writing, MCP provides a way to connect tools, data and services to AI models and systems. Through the Coralogix MCP Server, AI agents can directly access detailed information about a customer’s applications and infrastructure.
Last quarter, Coralogix introduced Olly, its AI observability assistant. Olly is a site reliability engineering (SRE) agent that can analyze production systems, understand the full context of logs, metrics and traces. It also helps surface root cause analysis (RCA) and business impact.
A Secure MCP Developer Endpoint
Onward from the chirpily-named Olly, the company says that the more formally-labelled Coralogix MCP Server is more aligned towards providing context to builders, programmers and software application developers. It works by offering a secure MCP endpoint so developers can stream live telemetry into their own AI agents, integrated development environments (IDEs), or chat-ops workflows to then help shape user experiences to suit.
“Agents generally lack direct access to specific observability data, which limits the AI’s utility for this purpose. What makes Coralogix’s MCP Server unique is its ability to surface observability data that is highly specific to each customer. It can search through data to find custom attributes and entities that reflect the customer’s unique setup, leading to more accurate results when AI agents access logs, metrics and traces. Customers can also use natural language prompts to locate key metrics or events,” said Coralogix’s Hason.
By integrating with tools developers already use, such as the widely used AI code editor Cursor or IDEs, the MCP Server enables AI agents to not only detect issues in real-time but also assist in diagnosing and resolving them all within the same workflow. Hason and team call this a “closing the loop capability” that streamlines operations and reduces the need to switch between multiple tools.
