Claude Code Plugins

Community-maintained marketplace

Feedback

Delegating to New Relic Agent

@ryancnelson/deli-gator
31
0

Recognize New Relic queries and delegate to specialized sub-agent to avoid context pollution

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name Delegating to New Relic Agent
description Recognize New Relic queries and delegate to specialized sub-agent to avoid context pollution
keywords newrelic, nrql, monitoring, alerts, apm, logs, drop-rules, observability, delegate, agent
when_to_use When user mentions "newrelic", "nrql", "alerts", "drop rules", or asks to query/search New Relic monitoring data. ALWAYS delegate to sub-agent.
version 1.0.0

Delegating to New Relic Agent

Core Principle

Never handle New Relic operations directly. Always delegate to a specialized sub-agent to keep your context clean and costs low.

Recognition Patterns

Delegate when user says:

  • "query newrelic for..."
  • "show me nrql results"
  • "list drop rules"
  • "any alerts firing?"
  • "search newrelic logs"
  • "what applications are monitored?"
  • "verify this nrql query"
  • Any mention of: newrelic, nrql, monitoring, alerts, apm, drop-rules, observability

How to Delegate

Use the Task tool with a specialized prompt:

Task(
  subagent_type: "general-purpose",
  description: "Query New Relic monitoring",
  prompt: "<full agent instructions from AGENT-INSTRUCTIONS.md>"
)

Agent Prompt Template

When delegating, include:

  1. The complete agent instructions (see AGENT-INSTRUCTIONS.md)
  2. The user's specific request
  3. Clear output format requirements

Example:

You are a New Relic monitoring specialist. Your job is to query New Relic using shell wrappers and return clean results.

<AGENT INSTRUCTIONS HERE>

USER REQUEST: Show me active alerts

Return a clean summary with:
- Alert severity
- Entity names
- Alert conditions
- Status

After Agent Returns

  1. Present results cleanly to user
  2. Offer follow-up if relevant (e.g., "Would you like to see logs for that service?")
  3. Don't expose mechanics (NerdGraph, GraphQL, auth, etc.) to user

Benefits

  • ✅ Main context stays clean
  • ✅ Cheaper queries (sub-agent uses less expensive model)
  • ✅ Specialized knowledge isolated
  • ✅ Scalable pattern for other monitoring services

Example Flow

User: "show me active alerts"

Main Assistant: [Recognizes New Relic query]
              → Invokes Task tool with agent instructions
              → Agent runs monitoring-alerts wrapper
              → Agent returns formatted results

Main Assistant: "You have 2 active alerts:

                CRITICAL: federated-graphql - High error rate
                WARNING: nginx-gateway - Response time elevated

                Would you like details on either of these?"

Common Query Types

NRQL Queries:

  • "query logs for errors"
  • "count transactions"
  • "average response time"

Drop Rules:

  • "show drop rules"
  • "list log filters"
  • "verify nrql before dropping"

Alerts:

  • "any alerts?"
  • "critical alerts"
  • "what's alerting?"

Applications:

  • "list monitored apps"
  • "find nginx service"

Logs:

  • "search logs for 502"
  • "nginx error logs"

Red Flags

DON'T:

  • ❌ Try to run monitoring-* scripts yourself
  • ❌ Construct NerdGraph GraphQL queries in main session
  • ❌ Load detailed New Relic API knowledge
  • ❌ Handle authentication directly
  • ❌ Parse complex monitoring data structures

DO:

  • ✅ Immediately delegate on New Relic keywords
  • ✅ Trust the sub-agent's results
  • ✅ Present clean summaries to user
  • ✅ Suggest relevant follow-up queries

Version History

  • 1.0.0 (2025-10-15): Initial delegation skill created to reduce context pollution