How to use SRE Command Center to manage incidents, monitor services, and automate remediation.
SRE Command Center is an AI-powered incident response platform that replaces the need to juggle multiple tools (Grafana, PagerDuty, Slack, GitHub, kubectl) during an outage. Instead, you interact with a single conversational interface that:
- Shows you what you need - Ask questions and get dynamic visualizations
- Explains what's happening - AI analyzes logs, metrics, and code changes
- Takes action on your behalf - Execute rollbacks, scale services, notify teams
- Learns from incidents - Store and analyze past incidents for patterns
- Navigate to
https://your-sre-command-center.com - Click "INITIATE_SESSION"
- Login with:
- Email/Password - If your admin set up email auth
- GitHub - One-click login with your GitHub account
- Google - One-click login with your Google account
After login, you'll be prompted to connect your infrastructure tools. These connections allow the AI to fetch real data.
| Service | What It Provides | How to Connect |
|---|---|---|
| Prometheus | Real-time metrics (CPU, memory, latency, errors) | Enter your Prometheus server URL |
| Kubernetes | Pod health, deployments, rollbacks | Upload kubeconfig or enter cluster URL |
| GitHub | Commit history, PRs, code changes | OAuth login to your GitHub org |
| Slack | Team discussions, incident channels | OAuth login to your Slack workspace |
| PagerDuty | Alert history, on-call schedules | API key from PagerDuty |
Note: Each connection is stored securely and scoped to your user account.
Click "INITIALIZE_COMMAND_CTR" to enter the main dashboard.
The core of SRE Command Center is the chat interface. You ask questions or give commands in natural language, and the AI responds with:
- Text explanations - Context about what's happening
- Dynamic visualizations - Charts, timelines, heatmaps
- Actionable controls - Buttons to execute remediations
| What You Ask | What Happens |
|---|---|
| "Show me the system status" | Displays a grid of all services with health indicators |
| "What alerts are firing right now?" | Shows a summary of active alerts by severity |
| "Why is the notification service failing?" | AI analyzes metrics, commits, and Slack to give root cause |
| "Show me what happened in the last hour" | Renders an incident timeline with events and metrics |
| "What services had anomalies today?" | Displays a heatmap of anomaly scores across time |
| "Rollback the notification service" | Shows remediation options with a confirmation button |
| "Copy the incident summary" | Copies formatted text to your clipboard |
| "Play an alert sound" | Plays an audible alert (for critical notifications) |
- Green border = Healthy
- Amber border = Degraded (elevated errors or latency)
- Red border = Critical (failing health checks)
Each card shows:
- Uptime percentage
- Current latency (ms)
- Error rate (%)
- Requests per second (RPS)
- Version number
When to use: Start here to get a quick overview of system health.
A grid showing anomaly scores for each service across time windows.
- Green cells = Normal behavior
- Yellow/Amber cells = Unusual patterns detected
- Red cells = Significant anomalies
When to use: Identify recurring issues (e.g., "payments fail every day at 4 PM").
A chronological view of:
- Alert triggers
- Automated actions (scaling, restarts)
- Human notes and acknowledgements
- Metric changes (error rate overlay)
When to use: During an active incident to track what's been done, or post-mortem to reconstruct events.
AI-generated analysis including:
- Suspected Cause - What the AI thinks is wrong
- Confidence Score - How sure the AI is (0-100%)
- Evidence - Specific signals that led to this conclusion
- Suspicious Commits - Code changes that correlate with the incident
- Recommended Action - What to do next
When to use: To accelerate the "what changed?" investigation phase.
When you ask the AI to fix something, it will show you available actions:
| Action | Risk Level | What It Does |
|---|---|---|
| Scale Horizontal | Low | Add more pods to handle load |
| Restart Pods | Low | Rolling restart of service pods |
| Enable Safe Mode | Low | Disable non-essential features |
| Rollback Deployment | Medium | Revert to previous known-good version |
| Traffic Shift | High | Move traffic to another region |
Each action shows:
- Estimated impact
- Step-by-step execution plan
- Confirmation button (you must click to execute)
Safety: High-risk actions require explicit confirmation and are logged.
SRE Command Center can send you browser notifications:
- Ask the AI: "Notify me if payment errors exceed 5%"
- Allow notifications when prompted by your browser
- You'll receive desktop alerts even if the tab is in the background
You can also:
- Play alert sounds for critical events
- Export incident data as JSON for external tools
- Copy summaries to paste into Slack or Jira
To connect Prometheus:
- Go to Settings > Integrations
- Click Add Prometheus
- Enter your Prometheus server URL (e.g.,
http://prometheus.monitoring:9090) - Click Test Connection
- Click Save
Once connected, asking "show me CPU usage" will query live Prometheus data.
To connect GitHub:
- Go to Settings > Integrations
- Click Connect GitHub
- Authorize the SRE Command Center GitHub App
- Select the repositories you want to analyze
Once connected, the AI can:
- List recent commits
- Identify breaking changes
- Link commits to incidents
To connect Kubernetes:
- Go to Settings > Integrations
- Click Add Kubernetes Cluster
- Either:
- Upload your
kubeconfigfile, OR - Enter the cluster API URL + service account token
- Upload your
- Click Test Connection
- Click Save
Once connected, the AI can:
- List pod health
- Show deployment history
- Execute rollbacks and restarts
After connecting your services, verify everything works:
Ask: "What's the current system status?"
Expected: A ServiceStatusGrid should appear showing your real services.
Ask: "Are there any active alerts?"
Expected: An AlertSummary should show alerts from your Prometheus/PagerDuty.
Ask: "What code changed in the last 24 hours?"
Expected: A list of commits from your connected GitHub repos.
Ask: "What remediation options are available for [service-name]?"
Expected: A list of actions like rollback, scale, restart.
- Ask: "Copy the incident summary"
- Paste into Slack, Jira, or email
- Ask: "Export the incident data as JSON"
- A file will download with:
- Timeline events
- Metrics snapshots
- Root cause analysis
- Actions taken
Before ending your shift:
- Ask: "Summarize what happened today"
- Copy and share with the next on-call engineer
- Check that your integrations are connected in Settings
- Verify network access to Prometheus/Kubernetes from SRE Command Center
- Check your permissions (do you have access to the cluster?)
- Check the cluster's health (is the API server responsive?)
- Be more specific in your question
- Ask: "Why did you conclude that?" to see the AI's reasoning
- All connections use TLS encryption
- OAuth tokens are stored encrypted in Supabase
- Actions are logged with user attribution
- Sensitive data (secrets, passwords) is never shown
Last Updated: 2026-02-07