Where are the results of background agents stored?

Background agents write their results to specific output files. When you launch a task, Claude returns an output_file path which you can access later using the Read tool or TaskOutput command.

How can I check the status of my background tasks?

You can list all active background tasks and their current status by typing '/tasks' in the terminal or using the 'Ctrl+T' shortcut to toggle the task list view.

How do I launch a task in the background using this skill?

You can launch a task in the background by adding 'run_in_background: true' to your Task function call within the Claude Code environment.

When should I avoid running agents in the background?

Avoid using background agents for tasks that take less than 10 seconds or for any operation whose output is required immediately for your very next step in the workflow.

Can I retrieve results from an agent before it finishes?

Yes, by using 'TaskOutput(task_id="...", block=false)', you can retrieve the current output of a running agent without waiting for it to complete.

Background Agents & Parallel Execution

Name: Background Agents & Parallel Execution
Author: SomtoUgeh

bySomtoUgeh

•

生产力与工作流

Enables non-blocking, parallel agent execution for long-running tasks such as multi-layered research and concurrent code reviews.

The Background Agents skill allows developers to maximize productivity within Claude Code by launching tasks asynchronously. By using the run_in_background parameter, users can trigger multiple specialized agents—such as code reviewers or research tools—simultaneously without locking their terminal session. This skill provides comprehensive patterns for monitoring task progress via terminal commands, retrieving partial or complete outputs, and managing complex workflows where multiple independent operations need to run in parallel while the developer continues active coding.

主要功能

01Real-time monitoring of all background tasks via /tasks and Ctrl+T

02Integrated file-based output tracking for persistent progress logs

03Parallel agent launching for simultaneous multi-tool execution

04Flexible result retrieval with blocking and non-blocking TaskOutput options

05Asynchronous task execution using the run_in_background parameter

062 GitHub stars

使用场景

01Executing non-gating pre-commit reviews or linting tasks while finishing active feature development

02Running multiple language-specific code reviewers in parallel during a PR review phase

03Launching deep codebase and external research agents while continuing to write documentation

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add somtougeh/somto-dev-toolkit background-agents

For use in Claude.ai and ChatGPT

Download Skill