Skip to main content
The Moose Builder is your primary interface for working with Wild Moose. Instead of learning YAML syntax, memorizing configuration options, or reading through documentation - just tell the Builder what you want to investigate and it handles the rest.
To get started immediately, just ask the Builder: “help me improve our autonomous-RCA coverage”. It will guide you through setting up and evaluating your own apparatus of debugging agents.

​
What the Builder does

The Builder is an AI agent that knows Wild Moose inside and out - the key concepts, the API, every integration, every configuration option. It acts as your expert guide, so you don’t have to memorize any of it.

Create agents conversationally

Describe what you want to investigate in plain language. The Builder translates your intent into fully configured debugging agents.

Test against real incidents

The Builder stress-tests your agents against historical issues to verify they find the right root causes before they go live.

Uncover tribal knowledge

If you have enough historical data, the Builder can infer investigation workflows automatically - surfacing the debugging patterns your team follows but may never have documented.

Iterate and refine

Review test results, adjust agent behavior, and re-test in a tight feedback loop. The Builder learns what works for your environment.

​
Meet you where you’re at

The Builder adapts to your starting point:
  • Rich historical data? The Builder analyzes your past incidents and proposes agents automatically, using the objective truth from hundreds of real events to build investigation workflows.
  • Some context but gaps? The Builder combines what it can infer from your tools with your team’s input to fill in the blanks.
  • Starting from scratch? Walk the Builder through your debugging process step by step, and it builds agents as you go.
Either way, agents are built as code - open, transparent, and fully configurable. Nothing is a black box.

​
Test-driven by design

Every agent the Builder creates goes through testing before it’s deployed:
  1. Historical replay - the Builder runs your agent against past incidents to verify it surfaces the right findings
  2. Test cases - each replay becomes a test case, so future changes don’t break what already works
  3. Confidence scoring - you see exactly how well an agent performs before it handles live alerts
This test-driven approach means your agents are reliable and predictable from day one.

​
Getting started with the Builder

The fastest way to start is to just ask:
  • “Help me set up RCA for our 5xx error alerts”
  • “What’s the best way to investigate high latency on our checkout service?”
  • “Can you analyze our last 2 weeks of incidents and suggest agents?”
The Builder will guide you from there. No prerequisites, no required reading.