incident.io MCP Server – Incident Response AI Access
incident.io provides an officially supported remote MCP server for connecting MCP-compatible AI assistants to operational incident data. Agents can query incidents, analyze alerts, check on-call coverage, manage escalations, and run operational analysis using the same permissions available through incident.io.
Overview
The incident.io MCP server is an officially supported remote Model Context Protocol server for teams that want AI assistants to work directly with their incident response data. It connects MCP-compatible clients such as Claude, ChatGPT, Cursor, Codex, and automation agents to incident.io without requiring a local server checkout or custom integration code.
The hosted server is designed for incident operations, on-call coordination, and post-incident analysis. It is currently described by incident.io as a Public Beta feature and can be enabled from the organisation's Settings → MCP area.
What the MCP server enables
The server exposes incident.io context and tools so an AI agent can answer operational questions and assist with response workflows. Documented use cases include querying incident history, analysing alerts, checking who is on call, managing escalations, and running deeper operational analysis.
Practical examples include asking how many high-severity incidents occurred in a time period, identifying which services or teams generated the most workload, reviewing overnight responder load, checking current on-call coverage, and analysing whether alerts became real incidents or noise. The documented incident_stats tool supports grouping incident data by dimensions such as severity, status, type, mode, alert source, custom fields, roles, and time periods.
When to use it
Use incident.io MCP when responders, SREs, engineering managers, or automation pipelines need structured access to live incident response knowledge from an AI tool. It is useful for handover preparation, incident trend analysis, alert-noise review, escalation context, and operational summaries that would otherwise require switching between incident.io dashboards, schedules, and reports.
Connection and authentication
The remote endpoint is https://mcp.incident.io/mcp using HTTP-based MCP transport. Interactive users authenticate through OAuth in their MCP client and access data as their own incident.io user account. Programmatic agents and internal automations can use an incident.io API key passed as a Bearer token in the Authorization header. API keys should be scoped to the minimum permissions required and rotated or deleted when no longer needed.
Key considerations
The hosted MCP server is available to paying incident.io customers while in Public Beta, so capabilities and tool coverage may change. OAuth is the right choice for human workflows because actions are attributed to a user and inherit dashboard-visible access. API keys are better for service actors and automation, but they do not expire until deleted, so they should be protected carefully. The earlier open-source local Go prototype was archived after incident.io released the officially supported remote MCP server.
Supported Transports
streamable_http
URL: https://mcp.incident.io/mcp
Frequently Asked Questions
- When should an AI agent use the incident.io MCP server?
- Use it when an AI assistant needs live incident response context from incident.io, such as incident history, alert analysis, current on-call coverage, escalation status, or operational summaries for handovers and reviews.
- What does the incident.io MCP server add to an AI agent's capabilities?
- It gives the agent structured access to incident.io data and tools instead of relying on static model knowledge. The agent can query incidents, analyse alerts, check on-call information, manage escalations, and summarize operational trends from the connected account.
- What can the AI agent access or manage through this server?
- Documented coverage includes incidents, alerts, on-call schedules, escalations, catalog-related context, workflows, and incident statistics. Actual access depends on the authenticated user's permissions or the scopes assigned to the API key used by the MCP client.
- How is authentication configured for this MCP server?
- Interactive users authenticate through incident.io OAuth from the MCP client, with actions attributed to their own user account. Automated systems can authenticate to https://mcp.incident.io/mcp using an incident.io API key in the Authorization header as a Bearer token.
- Which transport should be used for this MCP server?
- Use the hosted HTTP MCP endpoint at https://mcp.incident.io/mcp. The earlier local Go MCP server was a prototype and has been archived, so new integrations should use the officially supported remote server.