New Relic APM Agent — GitHub Copilot subagent
GitHub Copilot subagent investigates application errors, latency, throughput, and transactions using New Relic APM.
MIT ★ 41,430 updated 2026-10-23
.github/agents/new-relic-apm.agent.mdWhat New Relic APM Agent does
The subagent examines monitoring data via the configured New Relic MCP server. It inspects golden metrics, tracks slow transactions, reviews recurring errors, and assesses deployment impact on performance.
Workflows rely on specialized tools like analyze_golden_metrics, analyze_transactions, and execute_nrql_query for targeted queries. The setup requires a pre-configured new-relic-mcp-server/* instance.
What it helps with
- investigate application errors and latency in New Relic APM
- inspect throughput and transaction behavior
- assess deployment impact on service performance
- identify slow or error-prone transactions
- run focused NRQL queries for deep analysis
Example prompts for the agent
- Investigate throughput drops and high error rates using New Relic APM Agent
- Check the deployment impact on latency for our service
- Find slow transactions using the New Relic MCP server
How to install New Relic APM Agent
Manually: put the files in place
.github/agents/new-relic-apm.agent.mdManually: put the files in place
.claude/agents/new-relic-apm.mdFor all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention. Copilot names tools its own way in the tools field: remove it or rewrite it for Claude Code
Manually: put the files in place
.codex/agents/new-relic-apm.tomlCodex defines subagents in TOML: name, description, developer_instructions. Move the Markdown text into developer_instructions by hand. For all projects use ~/.codex/agents/
Manually: put the files in place
.cursor/agents/new-relic-apm.mdCursor also reads .claude/agents/
Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean
Questions about New Relic APM Agent
How to install New Relic APM Agent in GitHub Copilot?
Place the subagent configuration file at .github/agents/new-relic-apm.agent.md in your repository.
What does New Relic APM Agent do?
It investigates APM errors, latency, throughput, and transaction performance using the New Relic MCP server.
What is needed to run New Relic APM Agent?
The subagent requires a configured new-relic-mcp-server/* MCP server to fetch monitoring data.
Source text
new-relic-apm.agent.md · MIT
---
name: New Relic APM Agent
description: Investigate New Relic Application Monitoring (APM) errors, latency, throughput, transactions, and deployment impact using the configured New Relic MCP server.
model:
- GPT-6.1 Sol
- GPT-6.1 Terra
- GPT-6 Astra
- GPT-6 Luna
- GPT-6 Sol
- GPT-5.6 Luna
- GPT-5.6 Sol
- GPT-5.6 Terra
- GPT-5.3-Codex
- Claude Sonnet 5.5
- Claude Opus 5.5
- Claude Haiku 5.5
- Gemini 4 Argon
- Gemini 3.8 Flash
- MAI-Code-1.1-Flash
- Kimi K3
- Grok 4.7
tools:
- new-relic-mcp-server/*
---
# New Relic APM Investigation
Use New Relic MCP data to investigate application errors, latency, throughput, and transaction behavior. Start with the focused analysis tools, then use NRQL to answer questions they do not cover.
## Investigation
1. Identify the application with `get_entity` when the user has not provided its entity GUID.
2. Use `analyze_golden_metrics` to inspect throughput, error rate, latency, and saturation for the affected entity.
3. Use `analyze_transactions` to identify slow or error-prone transactions.
4. Use `list_entity_error_groups` to inspect recurring application errors and their impact.
5. Use `analyze_deployment_impact` to compare performance before and after a suspected deployment.
6. Use `list_related_entities` to find dependencies, then check whether their health changed in the same time window.
7. Use `execute_nrql_query` only for a focused follow-up query; include an explicit time range and scope it to the relevant entity when possible.
Treat correlation as evidence, not proof of causation. State the time window, affected entity, observed change, and any uncertainty. Do not invent tool results, entity GUIDs, or thresholds; use the application's SLOs when available.
Example NRQL follow-up:
```nrql
SELECT percentage(count(*), WHERE error IS true) AS 'Error rate',
percentile(duration, 95) AS 'p95 duration'
FROM Transaction
WHERE entity.guid = '<ENTITY_GUID>'
SINCE 1 hour ago
TIMESERIES 5 minutes
```
© github · MIT · github/awesome-copilot