Connect the incident.io MCP server to Unblocked to give AI agents live incident and on-call context
Unblocked connects to incident.io’s MCP server so AI coding agents can look up incidents, status updates, follow-ups, alerts, and on-call schedules live — then its Context Engine reconciles what they find with the code, pull requests, and conversations behind it.
incident.io integration details
What the incident.io integration adds
Before touching a fragile service, a developer should know whether it has paged someone recently, what the post-mortem concluded, and which follow-ups are still open. That history lives in incident.io, out of reach of the agent writing the change. incident.io’s MCP server exposes incidents, alerts, follow-ups, and on-call schedules to AI agents, and Unblocked connects to it once so every agent can ask without its own incident.io setup.
This is a third-party MCP server: incident.io hosts and maintains it, and Unblocked connects to it as a verified MCP server. Nothing is ingested or indexed — Unblocked calls the read-only tools you allow when a question needs them, so an incident’s status, its latest update, and who is on call come straight from incident.io at that moment.
Wiring incident.io’s MCP server straight into an agent gives it incident records. Unblocked’s Context Engine reasons across those records and your other sources — connecting an incident’s timeline to the pull request that caused it, the fix that followed, and the follow-up ticket that is still open — and returns one cited answer.
Once connected, any MCP-compatible coding agent — Claude Code, Cursor, Devin, GitHub Copilot, Codex — reaches incident.io context through a single Unblocked MCP server connection.
What you get
- Live incident history — Agents search incidents, read full incident details and status update history, and list post-incident follow-ups as they stand in incident.io right now.
- On-call and alert awareness — Agents look up on-call schedules and recent alerts, so they know who owns a service and whether it is already firing before suggesting a change.
- Read-only, tool-level control — Tools that declare, update, or merge incidents, resolve alerts, or create overrides are never offered. You choose which read-only incident.io tools Unblocked may call from Settings → Data Sources.
- Context Engine reasoning — incident.io records are reconciled with your code, pull requests, tickets, and conversations, so an incident leads to the change that caused it and the fix that closed it — with citations.
How to connect incident.io to Unblocked
Prerequisites
- An Unblocked account (free to start).
- An incident.io organization on the Team, Pro, or Enterprise plan, and an incident.io user who can authorize the OAuth connection. Some tools also need the Response or On-call product.
- An MCP-compatible coding agent (Claude Code, Cursor, Devin, etc.) to use the context in your editor.
Step 1: Connect incident.io to Unblocked
- In Unblocked, go to Settings → Data Sources → Connect another data source, and under Incident Management select incident.io.
- The incident.io MCP server URL is filled in for you. Choose OAuth, click Connect, and sign in to incident.io to authorize Unblocked.
- Once Unblocked discovers the server’s tools, select the read-only tools it may use and click Save Settings.
Step 2: Install the Unblocked MCP server
On macOS or Linux, one command auto-detects and configures the supported agents on your machine:
curl -fsSL https://getunblocked.com/install-mcp.sh | bash On Windows, or when you prefer remote MCP, add the server manually using the configuration below, or follow the manual install guide .
Claude Code
Add the remote server, then run /mcp in Claude Code, select unblocked, and log in via OAuth:
claude mcp add -s user --transport http unblocked https://getunblocked.com/api/mcpsseCursor
Open View: Open MCP Settings → Add Custom MCP, add the server, then click Connect to authenticate via OAuth:
{
"mcpServers": {
"unblocked": {
"url": "https://getunblocked.com/api/mcpsse"
}
}
}Other MCP clients
Any MCP-compatible client (Devin, Codex, VS Code, Claude Desktop, and others) can connect to the remote server with OAuth:
https://getunblocked.com/api/mcpsseincident.io MCP server tools
incident.io hosts and maintains this MCP server. Unblocked calls only the read-only tools you select, at the moment a question needs them:
incident_listSearches and browses incidents with filters and returns the matching incidents.incident_showReturns full incident details, optionally with the investigation, post-mortem, and update history.incident_update_listReturns the full status update history for an incident.follow_up_listLists post-incident follow-ups. Requires incident.io Response.schedule_listLists on-call schedules. Requires incident.io On-call.alert_listSearches and browses alerts and returns the matching alerts.
Connect incident.io to Claude Code, Cursor, and other AI agents
To give Claude Code, Cursor, GitHub Copilot, Codex, or another MCP-compatible AI coding agent access to incident.io context, connect incident.io to Unblocked and add the single Unblocked MCP server to your agent.
Unblocked acts as the MCP context layer for your incident.io data, searching incidents and incident updates alongside your other connected sources and returning cited results through these tools:
context_researchResearches a question across your connected sources and calls the incident.io tools you selected when incidents, follow-ups, alerts, or on-call schedules are relevant, returning a synthesized answer with citations.context_get_urlsFetches the full content behind URLs your agent encounters across connected sources — Jira, Linear, Notion, Confluence, Slack, GitHub, and more.
Questions to ask an AI agent about incident.io
Questions you (or your agent) can ask Unblocked about your incident.io data:
Has the billing service had any incidents this quarter, and which pull requests were linked to them?
What did the post-mortem for last week’s checkout outage conclude, and are its follow-ups done in Jira?
Who is on call for the payments team right now, and what alerts have fired for that service today?
Before I change this retry logic, was it involved in a past incident, and what did the team decide in Slack?