Get early access
Features · AI & Automation · Recipes
AI & Automation · Recipes

Playbooks as code,
runnable from anywhere.

A recipe is a versioned, parameterised sequence of commands. Run on schedule, run on alert, run on demand. Audited, dry-runnable, rollback-aware. The thing you keep meaning to extract from your runbooks.

recipe · restart-web v4
nginx -t
configuration file test is successful
systemctl reload nginx
curl -sf localhost/health
200 · recipe complete · audited
— Why this matters

The runbook in the wiki is not a runbook.

01

A command you can run is not a command you can trust.

A recipe is versioned and parameterised. What ran on Tuesday can be read back exactly, including which values were passed.

02

The same fix usually applies to more than one host.

One run modal covers a single server or fifty, so a fleet-wide fix is not fifty separate acts of copy-paste.

03

Everything a recipe does is auditable.

Runs land in history with their full output. "Somebody restarted something" stops being the answer.

— What's included

From writing one to running it everywhere.

Library and your own
System recipes to start from, plus whatever you write.
Multi-step editor
With syntax highlighting, because a runbook is code.
Parameters
Extracted and typed, so a recipe is reused rather than duplicated.
AI assistance
Generate a script from a description, optimise one you have, or pull the parameters out of an existing script.
Run on 1 to 50 hosts
The same modal either way, with per-server progress.
Bulk run overview
Every run lands on its own page, grouped when it spanned a fleet.
Scheduled
Cron-based, for the ones that should not wait for a human.
Full output history
Per step, per server, kept.
— How it works

Write once, run where it applies.

1

Write the steps

A multi-step editor with parameters, so the same recipe covers the variations instead of spawning copies.

2

Run on a selection

One host or fifty, with a bulk overview showing which succeeded and which did not.

3

Everything is auditable

Command, host, operator, exit code and output land in the audit log — that is what makes it a runbook rather than a shell history.

— Works with

Connected capabilities

Automations
Triggers × recipes — the wiring layer.
→
Auto-Recovery
Built-in recipes for the most common restart cases.
→
CLI Tool
Run any recipe from the terminal.
→
Audit Log
Where recipe invocations are recorded.
→
— FAQ

Questions, answered honestly.

Where do recipes live? +
Either in the platform or in your git repo (with a webhook for updates). Same execution either way.
Can a recipe call another recipe? +
Yes — recipes compose. Limited recursion depth, audit logged at every level.
How are secrets handled? +
Pulled from your secret store at run time; never persisted to the recipe definition. Vault, AWS SM, env-var integrations.

Turn your runbooks into runnables.

Recipes from Starter; concurrency + scheduled hooks from Pro.

Get early access Back to AI & Automation