Sentry vs New Relic
Sentry MCP and New Relic MCP are both official servers, but they reflect two different traditions of observability. Sentry is error-and-issue tracking first: its server lets an agent find organizations, teams, and projects, search issues, events, and releases, pull stack traces, manage DSNs and projects, and crucially run Seer — Sentry's AI root-cause analysis — to explain why an error is happening. It can run locally over stdio with an access token or remotely over OAuth at the hosted Sentry MCP endpoint. New Relic is full-stack APM and telemetry: its hosted server (remote over a bearer token) lets an agent run NRQL queries, fetch and list dashboards and entities, list related entities, analyze golden metrics, translate natural language into NRQL, and triage incidents across accounts. So one server is centered on debugging specific errors with AI root-cause help, while the other is centered on querying broad performance telemetry with NRQL. Here is a balanced look at how they differ.
How they compare
| Dimension | Sentry | New Relic |
|---|---|---|
| Core domain | Error and issue tracking: stack traces, events, releases, and DSNs, with Seer AI root-cause analysis for individual errors. | Full-stack APM and telemetry: NRQL queries, entities, golden metrics, and dashboards across your applications and infrastructure. |
| Query model | Search-oriented tools over issues, events, and releases, plus project and team management — framed around debugging. | NRQL-driven: execute_nrql_query plus natural_language_to_nrql, so the agent asks broad telemetry questions in New Relic's query language. |
| AI assistance | Built-in Seer root-cause analysis the agent can invoke to explain why a given error or issue is occurring. | Natural-language-to-NRQL translation and golden-metric analysis help the agent explore telemetry without hand-writing every query. |
| Hosting and auth | Runs locally over stdio with an access token, or remotely over OAuth at the hosted Sentry MCP endpoint. | Hosted remote server authenticated with a bearer token at New Relic's MCP endpoint. |
| Best-fit task | Debugging a specific exception — pulling its stack trace and events and running Seer to find the root cause, plus managing projects and DSNs. | Investigating overall application and infrastructure performance with NRQL, dashboards, entities, and incident triage. |
Verdict
Both are official observability servers, so choose by what you are investigating. Pick Sentry's server when your priority is errors and issues: you want an agent that searches issues and events, reads stack traces, manages projects and DSNs, and runs Seer for AI root-cause analysis on a specific failure. Pick New Relic's server when your priority is broad performance telemetry: NRQL queries (including natural-language-to-NRQL), entities, golden metrics, dashboards, and incident triage across your stack. In short: Sentry for error-centric debugging with AI root-cause; New Relic for full-stack APM querying with NRQL. Many teams run both — Sentry to catch and explain errors, New Relic to watch overall performance.
FAQ
- Does either server offer AI-assisted analysis?
- Yes, in different ways. Sentry exposes Seer, its AI root-cause analysis, so an agent can ask why a specific error is happening. New Relic offers natural-language-to-NRQL translation and golden-metric analysis, helping an agent query and interpret broad telemetry without hand-writing every NRQL statement.
- Can I run either locally?
- Sentry's server can run locally over stdio with an access token (and also remotely over OAuth). New Relic's server is hosted and remote-only, authenticated with a bearer token, so there is no local install for it.