# PagerDuty Incident Trigger MCP for AI Agents AI Agent Connect > PagerDuty Incident Trigger MCP lets your AI agent fire off incidents to your on-call team instantly. It's a surgical tool built for one purpose: notifying the right people when things break. No extra fluff, no complex auth, and no risk of the agent snooping on your data. It just works. ## Overview - **Category:** industry-titans - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_kKI8TivX0k47Sbfm98eG7QAfs3djX089vGi7fwVA/ai-agent-connect - **Tags:** incident-response, on-call, alerting, automation, events-api, site-reliability ## Description Most incident management integrations are bloated messes. They try to handle everything from on-call rotations to historical ticket querying, creating a massive security surface area for your production environment. We took a different approach. This Connector is a surgical, zero-trust bridge designed to do exactly one thing: trigger incidents via PagerDuty's Events API V2. When your AI agent spots an anomaly in your logs or detects a spike in error rates, it can immediately push a notification to the engineer on call. Because it's a one-way connection, your agent can't see who is on call, read old tickets, or mess with your schedules. It's the safest way to give your agent paging superpowers without opening the doors to your entire infrastructure. You'll find this in the Vinkius catalog as a lean, high-utility tool for teams who need to move fast without compromising security. It's built for those who want to automate the 'human-in-the-loop' part of incident detection without the headache of managing a full API integration. ## Tools ### trigger_pagerduty_incident Sends a new incident to PagerDuty with a summary, source, and severity level. It uses the Events API V2 to ensure your agent only has permission to push alerts. ## Prompt Examples **Prompt:** ``` The payment gateway is returning 500 errors. Trigger a critical PagerDuty incident. ``` **Response:** ``` I've successfully triggered a critical incident in PagerDuty for the payment gateway. The on-call team has been notified. ``` **Prompt:** ``` Alert the on-call team that the database latency is over 500ms. ``` **Response:** ``` Done. I've pushed a warning to PagerDuty regarding the database latency spike. They'll see the summary in their dashboard. ``` **Prompt:** ``` The staging environment is down. Send an info alert to PagerDuty. ``` **Response:** ``` Sent. I've pushed an info-level incident to PagerDuty for the staging environment. No immediate action is required, but the team is aware. ``` ## Capabilities ### Fire off PagerDuty incidents Your agent can immediately create a new incident when it detects a system failure. ### Set incident severity levels Specify if an alert is critical, an error, a warning, or just info. ### Identify the source of the issue Include the specific service or component that triggered the alert. ### Provide summary notes to engineers Attach a clear description of the problem so the on-call person knows what to check. ### Escalate critical alerts instantly Force a high-priority notification to ensure the right people are woken up. ## Use Cases ### Payment Gateway Failure An AI agent monitoring Stripe logs sees a 50% drop in successful transactions and uses trigger_pagerduty_incident to wake up the billing team. ### Database Latency Spike A monitoring agent detects slow queries and triggers a 'Warning' incident to let the DBA know before the site goes down. ### Security Breach Detection A security agent identifies an unauthorized login attempt and immediately pushes a 'Critical' alert to the security operations center. ### Deployment Error A CI/CD agent notices a failed production roll-out and notifies the engineering team via PagerDuty to initiate a rollback. ## Benefits - Give your agent the authority to declare a 'Critical' state using trigger_pagerduty_incident to ensure immediate response. - Reduce security risks because this Connector doesn't require broad OAuth scopes or access to your incident history. - Cut out the middleman by using a single Routing Key instead of managing complex SDKs or multi-step authentication. - Ensure engineers get clear context by including a summary and source directly in the trigger_pagerduty_incident call. - Maintain a strict zero-trust environment where the agent can't snoop on your on-call schedules or resolve existing tickets. ## How It Works The bottom line is your agent gets a direct line to your on-call rotation without any unnecessary permissions. 1. Add your PagerDuty Routing Key to the Connector configuration. 2. Point your AI agent at the system or data source you want it to monitor. 3. The agent calls the tool to push an alert the moment it detects a threshold breach. ## Frequently Asked Questions **Can the PagerDuty Incident Trigger MCP read my current on-call schedule?** No, this Connector is strictly one-way. It can only send new incidents to PagerDuty and cannot see who is currently on call or read any existing data. **How do I give my AI agent permission to trigger alerts?** You just need a single PagerDuty Routing Key. You add this to the Connector configuration, and your agent can start pushing alerts immediately. **Will this Connector create a new ticket every time it's called?** Yes, every time your agent calls trigger_pagerduty_incident, a new incident is created in PagerDuty based on the summary and severity you provide. **Is it safe to let an AI agent trigger PagerDuty incidents?** Yes, because this Connector uses the Events API V2. It doesn't have permissions to read, modify, or delete anything in your PagerDuty account. **Can I set the severity of the incident?** Yes, you can specify whether the alert is critical, an error, a warning, or info. It defaults to critical if you don't specify one. **Does this work with Claude and Cursor?** Yes, any MCP-compatible client like Claude, Cursor, or Windsurf can use this to interact with PagerDuty. **Can the agent resolve PagerDuty incidents with this?** No. This Connector exclusively uses the 'trigger' event action. It cannot resolve or acknowledge incidents. Its sole purpose is to wake up human operators when it detects a critical anomaly. **How do I get a PagerDuty Integration Key?** In PagerDuty, go to Services > Service Directory. Select your service, go to the Integrations tab, and add a new 'Events API V2' integration. Copy the resulting Integration Key.