> ## Documentation Index
> Fetch the complete documentation index at: https://metoro.io/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Assisted Debugging

> Use AI chat in Metoro to investigate production behavior, explore telemetry, and gather debugging context quickly

## Overview

Assisted Debugging gives you a chat-based way to work with Metoro's AI SRE layer.

Instead of manually clicking across logs, metrics, traces, Kubernetes state, and recent changes, you can ask Metoro a question directly and have it gather the relevant context for you.

This is useful when you want to:

* Ask a production question in plain language
* Investigate symptoms before you know the root cause
* Gather debugging context quickly across multiple signals
* Explore a service, cluster, or workload without building dashboards first

<Info>
  From anywhere in the platform, press <kbd>g</kbd> and then <kbd>c</kbd> to open Guardian Chat.
</Info>

## Ways to open Guardian Chat

<CardGroup cols={3}>
  <Card title="Keyboard Shortcut" icon="keyboard">
    From anywhere in Metoro, press `g` and then `c` to open Guardian Chat immediately.
  </Card>

  <Card title="Home Page" icon="house">
    Open the home page and use the Guardian prompt surface to start a new conversation.
  </Card>

  <Card title="Interactive Chat" icon="message">
    Navigate to **Guardian** → **Interactive Chat** to open the full chat experience directly.
  </Card>
</CardGroup>

## What you can ask

Assisted Debugging is best for exploratory questions such as:

* What is the most frequent error log in my cluster?
* Which service is using the most CPU right now?
* Find the slowest service and tell me which endpoint is slowest.
* Show me recent issues affecting a specific service.
* Check whether a deployment correlates with the current errors.

Because Metoro brings its own telemetry layer, chat can work across logs, metrics, traces, profiling, Kubernetes state, deployments, and code context when GitHub is connected.

## What Guardian Chat can do

<CardGroup cols={2}>
  <Card title="Logs" icon="file-lines">
    Ask Metoro to find relevant logs, error patterns, or the most common failure messages in your cluster or service.
  </Card>

  <Card title="Traces" icon="route">
    Inspect traces, latency bottlenecks, and slow endpoints without manually jumping between trace screens.
  </Card>

  <Card title="Graphs" icon="chart-line">
    Ask Metoro to plot graphs and gather metric context while you investigate a problem.
  </Card>

  <Card title="Investigations" icon="magnifying-glass">
    Investigate issues, deployments, alerts, and symptoms interactively until you reach a likely root cause.
  </Card>

  <Card title="Cluster Questions" icon="cube">
    Ask questions about your cluster, workloads, services, and infrastructure state in plain language.
  </Card>

  <Card title="Cross-Signal Debugging" icon="brain">
    Correlate logs, traces, metrics, Kubernetes state, and code context in one workflow.
  </Card>
</CardGroup>

## How to use it

<Steps>
  <Step title="Turn on AI SRE">
    Assisted Debugging requires AI SRE to be enabled for your account.
  </Step>

  <Step title="Open the chat surface">
    Open Guardian Chat using the `g` then `c` shortcut, from the home page, or from **Guardian** → **Interactive Chat**.
  </Step>

  <Step title="Ask a focused question">
    Ask a concrete question about a service, environment, alert, deployment, or symptom you are investigating.
  </Step>

  <Step title="Refine the investigation">
    Follow up with narrower questions to move from symptom to root cause, request more evidence, or ask Metoro to inspect a different signal.
  </Step>
</Steps>

## Returning to previous conversations

You can close the chat after asking questions and come back to it later.

<CardGroup cols={2}>
  <Card title="Guardian Agents" icon="bot">
    Go to **Guardian** → **Agents** and use the **Chat** filter in the top filtering menu to find previous chat-based runs.
  </Card>

  <Card title="Previous Conversations" icon="clock-rotate-left">
    Open a new chat and use the previous conversations side panel to jump back into earlier chats.
  </Card>
</CardGroup>

<Tip>
  Conversations persist after you close the chat, so you can resume an earlier debugging session instead of starting over.
</Tip>

## What makes it useful

* It reduces context switching across observability tools and screens.
* It can gather evidence from multiple telemetry types in one flow.
* It helps you get to a likely root cause faster when you are still forming the investigation.
* It works well alongside deployment verification, alert investigations, and autonomous issue detection.

## Best practices

* Ask about a specific service, environment, deployment, or time window when possible.
* Start broad if you are triaging, then narrow the follow-up questions quickly.
* Connect GitHub if you want code-aware answers and code-change correlation.
* Use AI alert investigations or autonomous issue detection when you want Metoro to investigate proactively rather than interactively.

## Related documentation

<CardGroup cols={2}>
  <Card title="AI SRE Overview" icon="brain" href="/ai-sre/overview">
    Understand how Assisted Debugging fits into Metoro's AI SRE workflows.
  </Card>

  <Card title="Alert Investigations" icon="bell" href="/ai-sre/alert-investigations">
    Investigate firing alerts automatically with AI.
  </Card>

  <Card title="Autonomous Issue Detection" icon="magnifying-glass" href="/ai-sre/anomaly-detection">
    Detect and investigate production issues without a preconfigured alert.
  </Card>

  <Card title="Code Fixes" icon="code" href="/ai-sre/github-integration">
    Add repository context and generated fixes to investigations.
  </Card>
</CardGroup>