Async Feedback Loop

Monitor FEEDBACK.md for mid-task user guidance during automation.

27|8|Updated Jan 16, 2026
One-click install
npx skills add https://github.com/Sounder25/Google-Antigravity-Skills-Library --skill async-feedback-loop
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: Async Feedback Loop
Source: https://github.com/Sounder25/Google-Antigravity-Skills-Library/tree/main/10_async_feedback
Command: npx skills add https://github.com/Sounder25/Google-Antigravity-Skills-Library --skill async-feedback-loop

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Long-running automation tasks often drift or require human adjustments. This skill provides a simple feedback channel via a FEEDBACK.md file so agents can pivot without restarting the task.

Core Features & Use Cases

  • Live feedback channel: Monitors FEEDBACK.md for new instructions and pivots behavior in real time.
  • Selective acknowledgement: Optionally marks feedback as read with a timestamp to maintain a clear feedback history.
  • Integration-ready: Designed for loop-based automation routines, dashboards, or agents that need quick user input mid-task.

Quick Start

Start the agent and ensure FEEDBACK.md exists in your working directory. In your task loop, check FEEDBACK.md for new instructions and pivot accordingly.

Frequently Asked Questions about Async Feedback Loop

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I add live user feedback to long-running PowerShell automation tasks?

No, a restart is not required. The skill monitors the FEEDBACK.md instruction channel to allow mid-task user guidance to be applied directly without restarting the workflow. It enables rapid adjustments to long-running automation tasks on the fly.

Can I use a custom instruction channel instead of FEEDBACK.md for mid-task pivots?

You can track read messages by using the optional --acknowledge flag. This logs reads with a timestamp in the FEEDBACK.md instruction channel, maintaining a clear feedback history for long-running automation tasks.

What is the best way to maintain a feedback history during loop-based automation routines?

A live feedback channel via a FEEDBACK.md file works by allowing agents to monitor the file for new instructions during a task loop. This enables mid-task pivots and rapid adjustments without restarting the workflow.

Do I need any external dependencies to monitor a markdown file for mid-task user guidance?

You can use the optional --file parameter to monitor a custom instruction channel instead of FEEDBACK.md. This allows mid-task user guidance and mid-task pivots to be applied seamlessly within your existing loop-based automation routines.