Help & Guides

A quick overview of what Cova can do. Each feature links to a detailed step-by-step guide below.

What is it?

Monitor Scan connects to your monitoring tools (PagerDuty, Datadog, Grafana, Sentry, New Relic, Sumo Logic, Splunk), pulls your live configuration, and runs a rule-based analysis that scores your monitoring coverage across multiple dimensions. When AI is enabled, it also generates prioritized recommendations and natural-language summaries.

What is it?

Repo Scan analyzes your actual codebase to discover endpoints, databases, message queues, and services - then cross-references them against your connected monitoring tools to identify what should be monitored but isn't. It bridges the gap between "what's configured" (Monitor Scan) and "what exists in code" (Repo Scan).

What is it?

Incident Autopilot is an AI-powered incident commander. Describe what's happening ("API latency spike on checkout service" or "users reporting 500 errors"), and it generates a structured investigation playbook pulling real-time data from all your connected tools - what to check first, which dashboards to open, who's on-call, and the blast radius.

What is it?

PR Guard automatically scans every pull request for new endpoints, databases, services, and message queues that need monitoring. It posts a comment directly on the PR with risk-scored findings, coverage status, and suggested monitor configurations - before the code gets merged. No AI activation required.

What is it?

Deploy Monitor lets you push AI-generated monitor configurations directly to your Datadog account from within Cova - no copy-pasting or manual setup required. When you click Generate Fix on a coverage gap or quality issue, Cova produces a ready-to-deploy config and gives you a one-click button to create or update monitors in Datadog.

What is it?

Ask Cova is an AI chat assistant that understands your monitoring configuration and (if scanned) your codebase architecture. It helps with incident triage, monitoring improvements, on-call questions, and architecture-aware troubleshooting.

New to Cova? Follow these steps to go from zero to your first monitoring analysis in about five minutes.

2

Navigate to Integrations

Click Integrations in the left sidebar. You'll see cards for each supported monitoring tool.

Integrations page

PagerDuty

Datadog

Grafana

Sentry

New Relic

Sumo Logic

Splunk

3

Connect your first monitoring tool

Click any tool card to expand it. Enter your API key or token, then click Connect. A green checkmark confirms the connection. See Connecting Tools for detailed credential instructions per tool.

Expanded tool card (example: PagerDuty)

PagerDuty Connected

API Key

••••••••••••••••

Connected

Want to see what Cova does before connecting your real tools? Demo mode loads sample data so you can explore every feature risk-free.

Cova analyzes your monitoring tools by connecting to their APIs. Each tool needs a specific credential. All credentials are encrypted at rest using AES-128 encryption.

Currently Supported Tools

More integrations are on the way. Today, Cova connects to:

Tool	Credential Needed	Where to Find It
PagerDuty	REST API Key (read-only)	PagerDuty → Integrations → API Access Keys → Create New API Key
Datadog	API Key + Application Key + Site/Region (includes write scopes for Deploy Monitor)	Datadog → Organization Settings → API Keys / Application Keys. Select your Datadog site from the dropdown (US1, US3, US5, EU, or AP1). See Scopes & Permissions
Grafana	Service Account Token (Viewer) + Instance URL	Grafana → Administration → Service Accounts → Add Token
Sentry	Auth Token + Organization Slug	Sentry → Settings → Developer Settings → Custom Integrations (scopes: project:read, org:read, event:read, alerts:read)
New Relic	User API Key (NRAK-xxx) + Account ID	New Relic → User menu → API Keys → Create a key (type: User)
Sumo Logic	Access ID + Access Key + Region	Sumo Logic → Administration → Security → Access Keys → Add Access Key
Splunk	Auth Token + Instance URL	Splunk → Settings → Tokens → New Token. Instance URL is your Splunk management endpoint (default port 8089, e.g. https://splunk.example.com:8089)

Coming Soon

The following integration is in development. Click the "Coming Soon" tool card in the app to request early access.

• Dynatrace - Software intelligence

How to Connect

2

Click a tool card to expand it

The card expands to show input fields for that tool's credentials.

Expanded tool card (example: Datadog)

Datadog

API Key

Paste your API key...

Application Key

Paste your application key...

Datadog Site

US1 (datadoghq.com) ▼

Options: US1, US3, US5, EU, AP1

Connect  ← Click

3

Enter your credentials and click Connect

Cova validates the credentials by making a test API call. If successful, a green checkmark appears and the card shows "Connected." If validation fails, you'll see an error message - double-check your key and try again.

Connected state

Datadog Connected

Disconnecting a Tool

To disconnect, expand the tool card and click Disconnect. An inline confirmation will appear - confirm to remove the credentials. You can reconnect at any time by entering new credentials.

Each tool requires specific API permissions for Cova to scan your configuration. Most tools only need read access. Datadog is the exception - write scopes are needed for Deploy Monitor (pushing AI-generated configs directly to Datadog).

Prerequisites

You need at least one monitoring tool connected (see Connecting Tools). The more tools you connect, the more comprehensive the analysis.

How to Run

1

Connect at least one tool

Make sure you have at least one monitoring tool connected (see Connecting Tools). Once connected, the Run Analysis button appears in the sidebar.

Sidebar with connected tool

Connected

PagerDuty

How Long Does It Take?

If the Analysis Fails

Each analysis is saved to your History tab (within Monitor Scan), so you can always go back and compare results over time.

After running an analysis, your dashboard displays a single scrollable view with everything you need. Here's what each section means.

Hero Row: Health Score + Coverage Score

The top of the dashboard shows two score rings side by side:

• Health Score (left) - Your overall monitoring health (0-100). Measures the total issues found across connected tools.
• Coverage Score (right) - How complete your monitoring setup is (0-100). Measures how well each area (alerting, escalation, etc.) is covered.

The Health Score is color-coded:

AI Summary + Recent Trend

If AI is enabled, an AI-generated narrative summarizes your monitoring posture with per-tool breakdowns. If you've run multiple analyses, a score delta shows whether your health score improved or declined.

Coverage Dimension Cards (with Findings)

The main body of the dashboard groups everything by coverage dimension - sorted worst-first so you focus on the biggest gaps. Each card shows a scored area of your monitoring with a tool logo, label, score bar, and percentage. Findings are embedded directly inside their related coverage cards, so you see issues in context.

Coverage scores follow the same color-coding as health scores:

Click a coverage card to expand it and see:

• Criteria - What Cova measures for this dimension
• AI recommendation - Specific guidance for improving this area (when AI is enabled)
• Issues - Related finding cards with severity, impact, action, status, and Generate Fix button
• Covered - Green-tagged list of items that have monitoring configured
• Missing - Red-tagged list of items that lack monitoring coverage
• Generate Fix - AI-powered button to create monitor configs for the gap. When multiple services are missing, generates one config per service. When quality warnings exist on covered monitors, generates improved configs for each warning

Which coverage dimensions appear depends on which tools you have connected:

Each area shows a percentage score. Scores start from a base (how much is covered) and can be penalized by related issues - for example, having 10 escalation policies but 5 of them misconfigured will reduce your Alert Quality score. Click any area to expand it and see what's covered, what's missing, and which specific issues are affecting the score.

What Cova Checks Per Tool

When you run an analysis, Cova inspects your tool configurations and flags issues at three severity levels: Critical (things that will cause missed incidents), Warning (risks that weaken your response), and Info (improvements worth considering).

When two or more tools are connected, filter buttons appear so you can view coverage from a specific tool.

Per-Tool Summary

Below the hero row, each connected tool gets a collapsible summary section. Click to expand and see:

This gives you a quick per-tool overview before diving into the full coverage dimension cards below.

Other Findings

Any findings that don't belong to a specific coverage dimension appear in an "Other Findings" section below the coverage cards.

Findings are the actionable output of each analysis. Cova lets you track their status, filter them, and export them.

Status Workflow

Changing Status

There are two ways to update a finding's status:

Filtering & Views

• Filter by severity - Click the severity pills (Critical, Warning, Info) in the filter bar to show only findings of that level
• Filter by tool - Click a tool pill to show findings from that tool only
• Filters apply everywhere - Active filters affect findings inside all coverage dimension cards simultaneously

Exporting a Report

Repo scanning lets Cova understand your actual codebase architecture and make context-aware monitoring recommendations - suggesting what should be monitored based on your code, not just what is configured.

Three Ways to Scan

Running a Scan

Automatic PR Scanning

If you've connected GitHub, enable PR Guard to automatically scan every pull request for new endpoints, databases, and services. Cova posts a GitHub comment flagging monitoring gaps directly on the PR.

Generating an Incident Runbook

After a scan, you can generate a runbook from the results - a document that combines your monitoring findings with architecture context for incident response. Look for the Generate Runbook option in the scan results view.

Ask Cova is an AI chat assistant that understands both your monitoring configuration and (if scanned) your codebase architecture. It can help with incident triage, monitoring improvements, and on-call questions.

When Is It Available?

Quick Prompts

When you open Ask Cova, you'll see prompt categories tailored to your setup. Four base categories are always available:

Contextual Prompts

Additional prompt categories appear automatically based on your connected tools and history:

Scoping to a Specific Tool

Use the scope pills above the input box to focus your question on a specific tool. When scoped, your question is automatically prefixed with the tool name so the AI focuses its answer on that tool's data. Click "All" to remove the scope.

Trend Questions

After running two or more analyses, you can ask Cova about trends. It has access to your last 5 analysis snapshots and will cite specific numbers in its answers - for example: "Your health score improved from 45 to 62 (+17 points). Critical findings dropped from 8 to 3."

Copying Responses

Each AI response has a copy button. Use it to paste answers into Slack, Jira tickets, or runbooks.

Tips for Getting Useful Answers

Incident Autopilot is an AI-powered incident commander that investigates production problems across all your connected tools. Describe what's happening and Cova generates a structured investigation playbook - what to check first, which dashboards to open, which logs to grep, who's on-call, and the blast radius.

Setup

1

Connect at least one monitoring tool

Go to Integrations and connect one or more tools (PagerDuty, Datadog, Grafana, Sentry, New Relic, Sumo Logic, or Splunk). The Autopilot pulls live data from all connected tools during an investigation - the more tools connected, the more comprehensive the results.

Integrations page

PagerDuty

Datadog

Grafana

Sentry

What to Type

Describe what users are experiencing, not what you think the technical cause is. Good examples:

What You Get Back

The investigation modal shows a structured playbook with 7 sections:

Service-Relevant Filtering

The Autopilot automatically filters results to prioritize items related to the affected services:

Download as PDF

Click the Download PDF button in the investigation modal to generate a branded document. The browser's print dialog opens, where you can save as PDF or print directly. The PDF includes all 7 sections formatted for sharing with your team.

Data Sources

The investigation pulls from all connected tools simultaneously:

PR Observability Guard automatically scans every pull request (GitHub) and merge request (GitLab) for new endpoints, databases, services, and message queues, then posts a comment flagging monitoring gaps - with risk scoring and suggested monitor configs. No AI activation required - it works using pure pattern matching.

What It Detects

Risk Scoring

Each detected endpoint is assigned a risk level based on what it does:

Each risk assessment includes a business-impact reason (e.g., "Unmonitored payment endpoints can cause silent revenue loss") so engineers understand why monitoring matters for that specific endpoint.

Setup - GitHub App Permissions

Enable PR Guard in Cova

What the PR Comment Looks Like

When a PR introduces new infrastructure, Cova's GitHub bot posts a comment with:

• Endpoints table - Risk level, HTTP method, path, file, and monitoring coverage status
• Coverage note - If no HTTP monitoring tool is connected, a note explains how to get accurate coverage checks
• "Why these endpoints need monitoring" - Expandable section with business-impact reasons for critical/high-risk endpoints
• "Suggested monitor configs" - Expandable section with ready-to-use JSON configs (Datadog, Grafana, or generic) with thresholds scaled by risk
• Database connections - New databases/ORMs found with recommended monitoring (connection pool, query latency)
• Message queues - New queues found with recommended monitoring (queue depth, consumer lag, dead letters)

Coverage Status

The "Status" column in the endpoints table adapts based on which monitoring tools you have connected to Cova:

Asterisked (*) statuses include a note under the table encouraging you to connect HTTP monitoring tools for more accurate findings.

Troubleshooting

GitLab Setup

For GitLab merge requests, you need to add a project webhook manually (GitLab doesn't have an "App" model like GitHub).

Connecting a source control platform lets Cova scan your repositories, power Incident Autopilot investigations, and enable PR Guard for automatic PR scanning. You can connect from Repo Scan, Incident Autopilot, or PR Guard - the steps are the same.

Connecting GitHub

GitHub uses a GitHub App installation flow. This grants Cova read access to your repositories and enables webhook-based PR scanning.

Connecting GitLab

GitLab uses an OAuth2 authorization flow. This grants Cova read access to your projects and the ability to post merge request comments for PR Guard.

Where You Can Connect From

GitHub and GitLab connections are shared across all features. Connect once, and it's available everywhere:

What Happens After Connection

• Your repository list loads automatically - visible on Repo Scan, Incident Autopilot, and PR Guard
• For GitHub, PR Guard webhooks are configured automatically via the GitHub App
• For GitLab, PR Guard webhooks need to be configured manually per project
• The connection persists across browser sessions via localStorage

What Happens After a Deploy or Restart

Disconnecting

To disconnect GitHub or GitLab, go to the Repo Scan page:

Setting Up Webhooks for PR Guard

Webhooks allow Cova to automatically scan pull requests and merge requests when they're opened or updated. GitHub and GitLab handle webhooks differently.

GitHub Webhooks (Automatic)

To verify webhooks are working:

GitLab Webhooks (Manual Setup)

GitLab doesn't have an "App" model like GitHub, so webhooks must be added to each project manually. This tells GitLab to notify Cova when merge requests are opened or updated.

Troubleshooting

Cova offers a free tier for evaluation and a Pro plan for teams that need unlimited access.

Free vs Pro

Understanding Usage Limits

Monthly limits (scans, chat, repo scans, integrations) reset on the 1st of each calendar month. Lifetime limits (Generate Fix, Runbook, Report Export, Incident Autopilot) are a total allocation that does not reset.

When you're approaching a limit, you'll see an "X left" badge on the relevant button. When a limit is reached, the button shows a limit message and you'll be prompted to upgrade.

View your current usage at any time from the Billing page in the sidebar.

Upgrading to Pro

Managing Your Subscription

Go to Billing in the sidebar and click "Manage Subscription". This opens the Stripe Customer Portal where you can:

• Update your payment method
• View and download past invoices
• Cancel your subscription

Invite teammates to your Cova workspace so they can access shared analysis results and monitoring insights.

Sending Invites

Accepting an Invite

Viewing Your Team

Accepted team members appear in the Settings page under the invite section. You can see who has joined and when they accepted.

How to Apply Generated Fix in Datadog

Cova generates Datadog monitor configs as JSON. If Datadog is connected, you can deploy directly from Cova. Otherwise, paste the config manually or use the API.

Option A: Deploy Monitor (one-click)

Option B: Paste in Datadog UI

Option C: Datadog API

How to Apply Generated Fix in Grafana

Cova generates Grafana alert rule or contact point configs as JSON. You can apply them through the UI or API.

Option A: Import via Grafana UI

Option B: Grafana API

How to Apply Generated Fix in PagerDuty

PagerDuty does not have a JSON import UI. Generated configs are applied via the PagerDuty REST API.

Creating a Service

Creating an Escalation Policy

Creating Event Orchestration Rules

How to Apply Generated Fix in Sentry

Sentry configs are applied via the Sentry REST API. There is no JSON import in the Sentry UI.

Creating an Alert Rule

How to Apply Generated Fix in New Relic

New Relic uses the NerdGraph GraphQL API to create alert conditions and notification destinations.

Creating an Alert Condition

Creating a Notification Destination

How to Apply Generated Fix in Sumo Logic

Sumo Logic configs are applied via the Sumo Logic REST API.

Creating a Monitor

Creating a Notification Connection

How to Apply Generated Fix in Splunk

Cova generates Splunk saved search configurations as JSON. You can apply them through the Splunk Web UI or the REST API.

Option A: Splunk Web UI

Option B: Splunk REST API

Use the Splunk management API to create saved searches programmatically:

How to Apply Generated Fix with Terraform

Every generated fix includes a Terraform tab with ready-to-use HCL code. This is the recommended approach for teams that manage infrastructure as code.

Supported Providers

Steps