New Relic Ground Truth feeds production data to AI coding agents via MCP

Closing the context gap between code and production

New Relic describes a familiar scenario: an alert fires for a rising error rate on a payment service while a developer is midway through a feature branch. The developer must leave the editor, find the right dashboard, write a New Relic Query Language (NRQL) query from memory and switch between traces and logs. According to the company, fifteen minutes can pass before the actual debugging begins.

New Relic Ground Truth is meant to remove that detour. The company says an AI agent already has the code, the branch and the recent commits, but knows nothing about how that code behaves once it runs. Ground Truth supplies telemetry, logs, traces, deployments and alerts through MCP, and New Relic says no custom integration work is required. All environments support both OAuth, which the company recommends, and API key authentication.

Plain-language questions instead of NRQL

New Relic lists example prompts that work across supported environments:

• "What are the top errors for my payment service in the last hour?"
• "Did my last deployment cause a performance regression?"
• "Give me p99 latency broken down by host for the last 30 minutes"

According to New Relic, Ground Truth automatically translates natural language into NRQL and runs the query through its MCP capabilities. The company says this removes the need to learn NRQL syntax and also helps engineers who do not use New Relic every day.

Use cases across Claude Code, Kiro and GitHub Copilot

Claude Code. With the New Relic MCP Server connected, a developer can ask "How did my last deployment perform?" from the terminal. New Relic says the server pulls together metrics, logs, traces, error rates and alert history into a deployment health report, and that Claude Code has enough context to begin work on a fix in the same session. The New Relic Claude plugin is live on the Anthropic marketplace and can be installed with claude plugin install newrelic@claude-plugins-official.

Kiro. New Relic launched a Kiro Power that pulls live telemetry, such as CPU bottlenecks and memory spikes, into the IDE. According to the company, Kiro can analyze that data and generate the infrastructure-as-code changes needed to right-size an environment.

GitHub Copilot. New Relic describes a workflow for finding gaps in observability coverage, such as missing APM, browser monitoring or change tracking. A GitHub Action tracks releases and triggers an analysis. GitHub Issues are then created for each gap, with an implementation checklist and links to documentation. The GitHub Copilot coding agent is assigned to each issue, implements the changes and opens a pull request. New Relic says future checks will extend to infrastructure monitoring, synthetics, logging and alerts.

GitHub custom agents. A custom agent tuned to the New Relic MCP Server allows short inputs to produce detailed answers. In the incident response example, a prompt such as "Check on the latest incident for this application" lets the agent create a full report and start a multi-step analysis of major incidents.

What it means for developers

New Relic argues that the scenarios are demonstrated in specific tools only because those teams happen to work there. The company says every workflow is available in every supported environment, and that because MCP is an open standard, new AI development environments inherit the capability as soon as they are created.

The company points to its 2026 AI Impact Report, which it says found a 5x improvement in teams' ability to ship features and updates to production. New Relic attributes this not to AI getting better at code but to AI gaining context about what the code does while running. The stated payoff is less context switching, faster root cause analysis and a shorter mean time to resolution (MTTR).