Skip to main content
You already have tools that watch your systems and hold your team’s knowledge. Connecting them means RubixKube hears about problems sooner, and Rubix has more to read when it works out why. There are two ways to connect a tool. An alert source sends RubixKube a webhook, and its events become incidents. A connection lets Rubix read a tool while it investigates and chats with you. Many tools can do both, and you set each one up separately.
Alert sources send webhooks into RubixKube and create incidents. Connections let Rubix read the tool. Do both for the best resultAlert sources send webhooks into RubixKube and create incidents. Connections let Rubix read the tool. Do both for the best result

Which one do you need?

  • You want Datadog alerts to become incidents. Set up an alert source.
  • You want Rubix to read Datadog logs in an RCA. Add a connection.
  • You want both. An RCA is only as good as what Rubix can read, so a webhook alone gives it less to work with.
  • You want Rubix to read your code or docs. GitHub, GitLab and Notion are connections.
  • Your tool isn’t listed, or it’s your own system. Use your own webhook to send alerts, or a custom integration so Rubix can read it.

Alert sources

An alert source is a tool that tells RubixKube when something breaks. It connects like any other source, under Environments > Connect. You pick the tool and RubixKube makes it a webhook of its own. You paste that URL into the tool. The source registers itself and shows up on Environments, and a built-in automation turns its events into incidents. Each tool’s page lists exactly which events open an incident and which resolve it.

Connections

A connection lets Rubix read a tool. It adds evidence to an RCA and answers to your questions in chat. For example, Rubix can look through Datadog logs for payments-api while it works out why the service is crash-looping.
Customize, Integrations page with tools in use and more to addCustomize, Integrations page with tools in use and more to add
An admin connects a tool under Customize > Integrations. Open a tool under Add more and sign in, or paste a token, depending on what the tool offers. Some tools ask for a site, region or customer id first. A few need an admin at the other tool to allow RubixKube before anyone can sign in, and the page tells you when that’s the case. See Sign-in and tokens. A connection’s Tools list shows everything the tool offers, with a switch for each one. During an RCA, Rubix only calls tools that read. In chat, any tool that changes something follows your modes.

What’s available in each

The catalog has the full list.

See what a connection can read

A connection that signs in doesn’t always reach anything useful. A token can have the wrong scope, or the logs can live somewhere else. So RubixKube checks. After you connect, sign in or press Test, RubixKube makes real reads against the tool over the last hour. The tool’s Reads section shows one row for each kind of evidence, such as Logs, Metrics, Alerts or Tickets, and each row says what happened. The same rows appear on an alert source’s page under Environments. Datadog, PagerDuty, Observe and Atlassian Rovo get a read for each kind of evidence. Other tools get one read that shows the connection answers.

Your own tools

To connect something that isn’t in the catalog, choose Add your own under Customize > Integrations. You can add an MCP server or a REST API. Custom integrations are Business and Enterprise.

Catalog

Everything you can connect.

Your own webhook

Send incidents from anything that can post JSON.

Slack

Talk to Rubix and follow incidents in Slack.

Rubix MCP

Use RubixKube from your AI tools.