Background Agents

Background agents are Alfred helpers that can run later, usually from a schedule or event. They are useful when you want Alfred to keep an eye on something or do a task without you opening the Dashboard at that moment.

When to use a background agent

Use a background agent for tasks like:

• Checking something on a schedule.
• Watching for a condition and reporting back.
• Creating a recurring workflow with more judgment than a simple reminder.

Use a normal reminder when you only need Alfred to notify you at a time or on a schedule.

Use a background agent when Alfred needs to do work or collect information before deciding whether to contact you and what information to include.

What background agents can do

A background agent can have instructions, keep working notes, and produce an inbox note for Alfred to show you later.

Alfred can provide background agents with card-access outside of their default juridstiction and can provide background agents with tools if Alfred deems it necassery for the background agent to function.

Usage examples

Here are some examples of when to use background agents, ask Alfred:

• “Every Friday morning, check the project notes and prepare a short status summary.”
• “Watch this page once a day and tell me if the price changes.”
• “Send me a rundown of tech news, weather forecast and neat quote every morning.”

Background agents are very complex and may require more handholding from you for more complex tasks. For example for tasks requiring multiple background agents and tiered progression you may want to be more precise with what you want. For example:

“Here’s the outline of a book I want you to write. Create a background agent for each chatper that’ll do its own independent research and write its chapter over time in a chapter.md file. Run 5 background agents at a time, each offset by 1 minute, with a cooldown of 5 minutes inbetween runs. Once a background agent completed its chapter, have it send me a message in Discord DMs. Once we’re through all chapters I’ll come back to you so you can stitch all the chapters together.”

Triggering, conditionals and enrichment

Alfred can subscribe background agents hooks which subscribe to specific events, once those events trigger, it’ll trigger the background agent for a single run. The most flexibile hook is Alfred’s scheduling hook, which allows Alfred to trigger agents on a schedule, but Alfred does have other hooks and may gain more hooks in the future.

Alfred can perscribe a pre-run script that’ll perform an action or run a check before potentially triggering the agent. This allows Alfred to automatically enrich the agent before each run or to run agents on a schedule where each run is dependent on a specific action. Here’s an example use-case:

“Monitor the Butler status page, every 10 minutes check whether there’s an active incident through a prerun script. If there is an active incident, provide the details to the background agent and have the background agent scan for tickets on the forum. Then provide a user sentiment summary and important threads that may need attention.”

Limits

Background agents are not a replacement for critical monitoring systems. Treat them as helpful automations, not guaranteed infrastructure.

They may be limited by your plan, usage allowance, available modules, and connected platforms.