> For the complete documentation index, see [llms.txt](https://docs.pagerly.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.pagerly.io/sre-ai-agent/remote-mcp-server.md).

# Remote MCP Server

Connect AI assistants like **Claude (Desktop & Code CLI)**, **Cursor**, **ChatGPT**, and automated AI coding agents directly to Pagerly using the Model Context Protocol (MCP).

***

## ⚡ Overview

The Pagerly Remote MCP Server exposes your on-call schedules, incident management, SRE AI investigations, MTTR metrics, alert noise analytics, and Status Pages directly to your local or cloud AI tools.

* **Server Endpoint:** `https://api.pagerly.io/mcp`
* **Protocol:** JSON-RPC 2.0 over HTTP / SSE
* **Authentication:** OAuth 2.0 (Browser Login) or API Key (`Authorization: Bearer <token>`)

***

## 🛠️ Available MCP Tools

### 1. `pagerly_get_oncall`

Query who is currently on call for any team or rotation schedule across PagerDuty, Opsgenie, Datadog, Google Calendar, and Outlook.

```json
{
  "org_id": "T0123456",
  "team_name": "Platform"
}
```

### 2. `pagerly_create_incident`

Declare a new incident, provision a dedicated Slack channel, assign severity (P1-P4), and set the incident commander.

```json
{
  "org_id": "T0123456",
  "title": "High Latency on Payment Gateway",
  "severity": "P1",
  "service_id": "payments-service"
}
```

### 3. `pagerly_update_incident`

Update ongoing incident status (*Open, In Progress, Investigating, Identified, Monitoring, Resolved*) or severity.

```json
{
  "org_id": "T0123456",
  "incident_id": "INC-102",
  "status": "Resolved"
}
```

### 4. `pagerly_get_ai_investigation`

Fetch or trigger a deep SRE AI investigation for an incident. Returns:

* **Telemetry:** Datadog / CloudWatch Metrics (-15m) & Error Logs (300 lines)
* **Deployments:** Recent GitHub deploy events (-2h)
* **Root Cause Hypotheses & Debugging Next Steps**

```json
{
  "org_id": "T0123456",
  "incident_id": "INC-102",
  "run_new": true
}
```

### 5. `pagerly_incident_stats`

Query incident analytics, Mean Time to Resolve (MTTR), severity distribution, and recurring root-cause hypothesis themes.

```json
{
  "org_id": "T0123456",
  "days": 30
}
```

### 6. `pagerly_alert_noise_analysis`

Analyze alert volume vs actual incidents, noise ratio %, top noisy alert sources (CloudWatch, Datadog, etc.), and off-hours page percentages.

```json
{
  "org_id": "T0123456",
  "days": 30
}
```

### 7. `pagerly_update_statuspage`

Update component health (*Operational, Degraded Performance, Partial Outage, Major Outage*) or publish public status updates.

```json
{
  "org_id": "T0123456",
  "component_name": "API Gateway",
  "status": "Degraded Performance",
  "incident_message": "Investigating elevated error rates."
}
```

### 8. `pagerly_update_schedule`

Update team rotation schedule ID, rotation frequency (*daily, weekly*), or timezone settings.

```json
{
  "org_id": "T0123456",
  "team_name": "Platform",
  "rotation_frequency": "weekly",
  "timezone": "America/New_York"
}
```

### 9. `pagerly_set_reminder`

Set or update shift rotation reminder notifications for an on-call team.

```json
{
  "org_id": "T0123456",
  "team_name": "Platform",
  "reminder_time": "1day"
}
```

***

## 🔑 Quick Setup

### 1. Claude Code CLI

Add the server via CLI:

```bash
claude mcp add pagerly --transport http https://api.pagerly.io/mcp
```

### 2. Claude Desktop & Cursor

Add to your `claude_desktop_config.json` or Cursor MCP settings:

### 3. Claude.ai Custom Connectors (Enterprise)

If Claude prompts for manual OAuth settings in custom connector options:

* **Server URL:** `https://api.pagerly.io/mcp`
* **Authorization URL:** `https://workspace.pagerly.io/oauth/authorize`
* **Token URL:** `https://workspace.pagerly.io/oauth/token`
* **OAuth Client ID:** `pagerly_mcp_client` *(or auto-registered via `.well-known` discovery)* Upon first query, Claude will prompt you to authenticate via OAuth browser login.