Creating a Runbook

Route: /rb/new · Edition: Pro+

The runbook builder lets you define a named, ordered sequence of steps. Each step has a type and a type-specific configuration.

Basic fields

Field	Description
Name	Unique runbook name (required)
Description	What this runbook does
Tags	Comma-separated tags for filtering and AI matching

Triggers

Runbooks can be triggered manually, from the AI Chat, or via a webhook URL. Add a trigger to expose a POST /api/runbooks/:id/webhook endpoint with a secret token.

Steps

Each step has:

• Order — execution sequence (drag to reorder)
• Name — label shown in execution output
• Type — one of the supported step types below

Step types

Type	What it does
HTTP	Make an HTTP request (GET/POST/PUT/DELETE/PATCH) to any URL
Jira	Create an issue, add a comment, or search tickets
Slack	Post a message to a Slack channel
Shell	Run a shell command on the server
MCP	Call a tool on a configured MCP server
Condition	Branch execution based on a previous step's output

HTTP step config

Method:          GET | POST | PUT | DELETE | PATCH
URL:             https://api.example.com/health
Headers:         { "Authorization": "Bearer {{token}}" }
Body:            (JSON, for POST/PUT)
Timeout:         30 seconds
Expected status: 200

Shell step config

Command:  kubectl get pods -n production
Timeout:  60 seconds
Env:      { "KUBECONFIG": "/etc/kubeconfig" }

Safety validation

After adding or changing any step, the platform re-evaluates the runbook's risk level and shows a coloured alert. For Dangerous runbooks you must acknowledge the risk before saving.