NewRelicAgentObservability.md

Challenge 1: Foundations Reading Guide (Public Gist)

Challenge 1 is intentionally a docs-first challenge. Read these in order, then move to coding.

What You Need To Understand

1. Agent vs. plain LLM call
2. How ChatAgent uses tools
3. Why tool functions need good type hints + docstrings
4. How Flask request flow connects to agent execution
5. Why traces, metrics, and logs are essential for AI debugging

Read These Docs First (In Order)

1. Agent Framework Overview Why: Core concepts and vocabulary used in the rest of the hack.

2. ChatAgent tutorial section Why: Main programming abstraction used in later challenges.

3. OpenTelemetry Concepts Why: Defines trace/span/metric/log fundamentals.

4. Why Observability Matters (New Relic) Why: Practical production value of observability.

Recommended Next

1. Microsoft Agent Framework GitHub Why: Real examples and implementation patterns.

2. Flask Quickstart Why: Refresh API-layer basics used by the app.

5-Minute Self-Check

You are ready for Challenge 2 if you can answer:

1. What can an agent do that a single LLM call cannot?
2. Why does the agent need tools?
3. What does a trace show that logs alone do not?
4. If latency spikes, how do traces + metrics help isolate cause?
5. Where does Flask stop and the agent runtime begin?

Definition Of Done

Proceed when you can:

• Explain the agent-tool lifecycle clearly
• Give one practical example each for trace/span/metric/log
• Name 3 telemetry signals you would monitor in an AI travel planner