Playbook

Field	Description	Scheme
description	A short description	string
title	Title shown on the UI. Defaults to the playbook's name	string
icon	Icon for the playbook	Icon
category	Group the playbook under a category on the UI	string
timeout	Maximum duration to let the playbook run before cancellation. Valid time units are "s", "m", "h", "d", "w", "y". Defaults to 30 minutes.	string
on.canary	Run a playbook when a health check fails or passes	EventTrigger
on.component	Run a playbook when a part becomes heathy/unhealthy	EventTrigger
on.config	Run a playbook when someone creates/updates/deletes a config item or changes its state	EventTrigger
on.webhook	Run a playbook when someone calls a webhook	Webhook
on.schedule	Run a playbook on a recurring cron schedule	[]Schedule
runsOn	Which runner (agent) to run the playbook on	[]Agent
templatesOn	Where the templating of actions occurs For host the templating occurs on the mission control instance before sending to the agent For agent the templating occurs on the agent/runner where there might be secrets not accessible by the primary instance.	host or agent
checks	Which health checks this playbook can run on	[]ResourceSelector
configs	Which config items this playbook can run on	[]ResourceSelector
components	Which parts this playbook can run on	[]ResourceSelector
filters	CEL expressions that decide whether the playbook can run on the selected resource	[]Expression
env	Variables to lookup, available as env map in templating/filters	[]EnvVar
parameters	Variables that users need to enter. Do not use parameters for sensitive values.	[]Parameter
jsonSchema	A JSON schema, or a URL to one, to use for the run form instead of parameters	string
ui	Properties applied to the UI form	map[string]any
actions	Individual actions or steps to perform	[]Action
approval	Optional approvals required before a playbook runs	Approval
permissions	Roles and teams that are allowed to run this playbook	[]Permission
mcp	How this playbook is presented to LLM clients as an MCP tool	MCP

Run

Field	Description	Scheme
agent_id	ID of the agent that executed the playbook run	string
check_id	ID of the check associated with the playbook run	string
component_id	ID of the component associated with the playbook run	string
config_id	ID of the config associated with the playbook run	string
created_by	ID of the user who created the playbook run	string
id	ID of the playbook run	string
playbook_id	ID of the playbook	string
status	Status of the playbook run	string

Actions

Field	Description	Scheme	Required
name	Step Name	string	true
runsOn	Which runner (agent) to run the action on	[]Agent
templatesOn	Where templating (and secret management) of actions occurs	host or agent
delay	A delay before running the action e.g. 8h	Duration or CEL with Playbook Context
if	Conditionally run an action	CEL with Playbook Context
timeout	Timeout on this action.	Duration
retry	Retry the action when it fails	Retry
contentType	How the action's primary output is rendered on the UI	text/plain, text/markdown, text/x-shellscript, application/json, application/yaml, application/log+json or application/sql
ai	Prompt an LLM with the context of the resource	AI
azureDevopsPipeline	Trigger a pipeline run	AzureDevops
catalog	Create a config item in the catalog	Catalog
exec	Run a script e.g. to use kubectl or aws CLIs	Exec
github	Trigger Github Action	Github Action
gitops	Update a git repository (directly or via pull request)	Gitops
http	Call an HTTP Endpoint	Http
logs	Fetch logs from Loki, CloudWatch, OpenSearch or Kubernetes	Logs
notification	Specify notification of action.	Notification
pod	Run a kubernetes pod.	Pod
prometheus	Run a PromQL query	PrometheusQuery
report	Render a catalog report from a view or a config selector	Report
sql	Execute a SQL query	Sql

Only 1 action should be specified per step

Retry

Field	Description	Scheme
duration*	Duration to wait before retrying the action	Duration
exponent.multiplier*	Exponential backoff multiplier applied to the duration on every retry	integer
limit*	Number of times to retry the action. With a limit of 3 there is a max of 4 attempts (initial attempt + 3 retries)	integer
jitter	Random factor, from 0 to 100, applied to the wait duration	integer

Approvals

Approvals allow requiring one or more people to approve before a playbook runs.

scale-deployment.yaml
apiVersion: mission-control.flanksource.com/v1
kind: Playbook
metadata:
  name: delete-pv
spec:
  description: Delete Persistent Volume
  configs:
    - types:
        - Kubernetes::PersistentVolume
  approval:
    type: any
    approvers:
      teams:
        - DevOps
  actions:
    - name: kubectl delete pv
      exec:
        script: kubectl delete persistentvolume {{.config.name}}

Field	Description	Scheme	Required
type	How many approvals required. Defaults to all	any or all	false
approvers.[]people	Login or id of a person	People	false
approvers.[]teams	Name or id of a team	Team	false

Permissions

Permissions grant roles and teams the ability to run this playbook, in addition to the global permissions.

Field	Description	Scheme	Required
role	Name of a role	string	false
team	Name of a team	string	false
ref	Name of a Permission resource	string	false

MCP

Mission Control exposes playbooks to LLM clients as MCP tools. The mcp field controls how a playbook is presented to those clients.

Field	Description	Scheme	Required
title	Tool title shown to the LLM. Defaults to the playbook's title	string	false
description	Additional context for the LLM, beyond spec.description	string	false
tags	Keywords used to categorize the tool for LLM discovery	[]string	false
readOnlyHint	The playbook does not modify any state	bool	false
destructiveHint	The playbook may perform destructive operations	bool	false
idempotentHint	Repeated runs with the same arguments have no additional effect	bool	false
openWorldHint	The playbook interacts with entities outside Mission Control	bool	false