Background Agents

Set up agents that do work for you in the background, preparing investigations on incoming issues before you arrive.

This feature is currently in beta, with access limited to those participating in the beta.

Background Agent runs asynchronous workflows so work can happen in the background.

For example, it can investigate a new issue, gather evidence, run a Skill, and prepare findings for Assist Agent.

How Background Agent works

Background agents can be configured from the Agents page:

1. A configured trigger matches an issue, or an external system sends a webhook.

2. Background Agent runs the instructions provided using the context, tools, and skills available to it.

3. Depending on it's configuration it either sends its results into the Assist Agent panel or runs fully in the background.

4. A teammate reviews the output with Assist Agent and can ask follow-up questions or take further action.

Example: Issue Investigator

A background agent prepares an investigation each time a new issue arrives.

Example: Update documentation

A background agent that runs when there is a new feature released to update relevant knowledge base articles.

Context

You have full control over which tools, skills, and connectors the Background Agent has access to.

Depending on your configuration, Background Agent can work with:

• The context of the triggering object (issue, account, webhook, etc)

• Knowledge base articles and connected documentation

• Past issues

• Account fields and notebooks

• Connected codebases and logs

• MCP connectors and other enabled tools

• Skills assigned to the agent

For any connectors that are enabled for the background agent, they will use the Agent connection - native tools and tools from connectors can be enabled or disabled on the Tools tab of the Agent.