Automations

Automate recurring documentation work with automations

Automations run AI-powered documentation tasks in the background so you can catch issues, sync docs from code, and generate updates without repeating the same manual process.

What automations are

Automations handle common documentation jobs inside Documentation.AI. You can use them to audit content, enforce standards, generate updates, and run repo-aware tasks that use your connected documentation and code context.

Each automation starts from a template or a custom setup. After you create it, the automation can run on a trigger such as a pull request merge or a schedule, and you can also run it manually when you need an on-demand result.

Where to find automations

In the dashboard sidebar, under AI, click Automations. The Automations page is where you create automations, review previous runs, and inspect results.

Before you start

Automations depend on your plan and repository connection setup. Confirm these requirements before you try to create one.

If Documentation.AI can't reach the repository your documentation is published from, the Automations page shows We cannot reach your repository instead of the templates. Click Fix connection to open your Git settings.

Built-in automation templates

Start with a built-in template when you want a focused automation for a common task. Choose Start from scratch under Custom when none of the templates matches your process.

Templates are organized into two categories in the template picker:

• Self-updating templates generate or update documentation content based on code changes, repository activity, or user feedback.
• Auditing templates check and report on documentation health without modifying content directly.

Self-updating templates

Auditing templates

The best template depends on the job you want to automate. Content-only checks such as style or navigation audits usually require less setup than repo-aware automations such as code-driven updates or API sync tasks.

Create an automation

Create your first automation from the Automations page in a few steps. The setup flow asks you to choose a template, define how it should run, and pick a trigger repository when needed.

Triggers and repository context

Triggers control when an automation runs. Connectors control what context the automation can read.

Trigger options

Use the trigger that matches how often the task should happen:

• On pull request merge for automations that should react to repository changes

• On schedule for recurring checks such as audits or maintenance tasks

To start a run on demand, click Run now (the play icon) next to the automation in the automations list, or the play button on its edit page.

If you schedule an automation monthly, pick a day from 1 through 28. Monthly scheduling does not support days later in the month.

Trigger repository and code context

The trigger repository is the one repository an On pull request merge automation watches. A merge in that repository starts a run.

Code context is separate from the trigger. Whenever GitHub is selected under Connectors, the agent reads code from the context repositories you linked in the GitHub connector, the same ones it uses in the editor.

Review automation runs

Automation runs happen in the background. You can leave the page and return later to check progress or inspect the result.

Run history shows the state of each run as it moves through the automation lifecycle. You may see statuses such as queued, planning, executing, validating, committing, completed, human needed, no action needed, failed, and cancelled.

When a run finishes, the result can include:

• A summary of what the automation found or changed

• A View changes button, when the run changed files, that opens the run's branch in the editor with its changes shown

• A Chat with Agent button, when a Broken Link Audit run has findings, that opens the editor chat pre-loaded with them so you can ask follow-up questions or request fixes

Those outputs help you decide whether to accept the result, investigate further, or continue the task in the editor chat.

Credit usage

Automations consume AI credits. Credit usage depends on the type of automation and the amount of work required to complete it.

Simple checks usually use fewer credits than repo-aware or multi-step automations. If an automation analyzes multiple repositories or performs more planning and validation, expect higher credit usage.

Current limitations

The current Automations experience covers the most common user-facing automation paths, but a few constraints affect setup and scheduling today.