Product / Remote MCP

One remote MCP endpoint for production context.

IncidentFlow runs as a remote MCP server your AI client connects to over HTTP. There is no local process to install or keep running — add one endpoint, authorize in the browser, and your assistant can investigate production through traceable tool calls.

Connect IncidentFlowRead the docs

Connection path

  1. 01AI client
  2. 02Remote MCP endpoint
  3. 03Workspace
  4. 04Production integrations

Why remote

A hosted connection your whole team can rely on.

01

Nothing to run locally

The MCP server is hosted and reached over HTTP. Your client points at one URL instead of installing and maintaining a local process.

02

One endpoint, every client

Codex, Claude Code, and other supported clients connect through the same remote endpoint, so the connection behaves the same everywhere.

03

Traceable by design

Every action is an explicit MCP tool call. An investigation stays legible as a sequence of steps, evidence, and results.

How it connects

Add one URL. Authorize once. Investigate.

  1. 01

    Add the server

    Register the IncidentFlow remote MCP URL in a supported AI client.

  2. 02

    Authorize in the browser

    Complete browser-based authorization, then choose the workspace intended for the work.

  3. 03

    Investigate production

    Ask about cluster health, recent changes, or telemetry — answered through live tool calls.

codex mcp add incidentflow --url https://mcp.incidentflow.io/mcp

The remote endpoint stays consistent across clients, so there is no local MCP process to maintain and no per-machine setup to keep in sync.