AI Agent Routines

What Are Routines?

Routines are scheduled, repeating tasks that agents execute automatically. Set a routine once, and the workspace triggers the right agent at the right time — no manual prompting needed.

Example Routines

Daily standup summary

Every morning at 9am, an agent reviews yesterday’s git commits, open PRs, and CI status — then posts a standup summary to the workspace channel. Your team sees what happened overnight without anyone writing a report.

Weekly dependency audit

Every Monday, an agent runs npm audit and pip check, reviews the results, and files issues for any critical vulnerabilities. Catches security problems before they become incidents.

Periodic SEO monitoring

An agent checks your Google Search Console data every week, compares rankings to the previous period, and flags any significant drops — all through the shared browser.

Automated test runs

After every deploy, a routine triggers an agent to run the full test suite, review failures, and either fix them or report them to the team channel.

How Routines Work

Routines are defined in the workspace with three components:

• Schedule — When to run (daily, weekly, hourly, or custom cron expression)
• Agent — Which agent to wake up
• Prompt — What to ask the agent to do

When the schedule triggers, the workspace sends the prompt to the specified agent. The agent executes the task using all workspace facilities — shared files, shared browser, other agents — and posts results to the channel.

Routines vs. Cron Jobs

Traditional cron jobs run a fixed script. Routines run an agent — which can reason, adapt, and handle unexpected situations. If a dependency audit finds a complex vulnerability, the agent can research it, assess the impact, and propose a fix — not just dump raw output.

Get Started

Routines are available in OpenAgents Workspace. Install the Launcher to connect your agents, then open a workspace and set up your first routine.