# Raygun for AI Agents

Connect AI coding agents to Raygun remote MCP and API V3 so they can investigate production errors, sessions, deployments, and traces with evidence.

## Purpose

Raygun is an observability source for coding agents. Its Crash Reporting, Real User Monitoring, and Application Performance Monitoring data can provide production evidence for investigation and repair work.

## Available production context

- Crash Reporting: error groups, stack traces, affected users, breadcrumbs, custom data, deployments, and ownership context.
- Real User Monitoring: sessions, browser and device context, route timing, geography, user journeys, and Core Web Vitals.
- Application Performance Monitoring: traces and dependency context for separating backend bottlenecks from client-side symptoms.

## Evidence-first investigation loop

1. Scope the work with an error URL, deployment version, customer, or route.
2. Use Raygun MCP to collect stack, session, deployment, and impact evidence.
3. Propose the smallest code change and a regression test.
4. Have a human review the resulting change before merge or deployment.

## Integration choices

- Use [Raygun remote MCP](https://ai.raygun.com/mcp) for interactive coding-agent investigations.
- Use [Raygun API V3](https://ai.raygun.com/api) for deterministic scripts, CI checks, and custom integrations.
- Follow the [setup guide](https://ai.raygun.com/get-started) before asking agents to investigate production data.