Build a physical status light indicator for Claude Code agentic runs
An engineering hack uses a DIY status light to track when Claude Code is executing tasks or waiting for input. This physical feedback loop helps developers monitor long-running background agent loops without constantly watching the terminal.
Impact: Low
Why it matters
It provides an ambient physical interface to track background AI agents, preventing context switching while waiting for long generation steps.
TL;DR
- 01Physical ambient notifications reduce the cognitive load of monitoring background agent runs.
- 02Using wrapper scripts or stream parsing allows DIY hardware integration even without native APIs.
Physical Feedback for Agentic Loops
CLI-based coding agents often execute complex, multi-step tasks in the background. A physical status light acts as an ambient notifier, flashing or changing color depending on whether the agent is actively writing code, running tests, or waiting for user confirmation.
Implementation Strategy
Without official API webhooks for Claude Code's CLI state, developers typically implement this by wrapping the terminal command or monitoring active processes. A background daemon can check for CPU usage spikes of the runner or parse the output stream, then trigger a local API call to an ESP8266 or Arduino-based LED strip.
Try it in 2 minutes
# Example of wrapping a CLI process to toggle a local smart light or LED controller via curl
claude code --write "implement auth" | while read -r line; do
if echo "$line" | grep -q "Thinking"; then
curl -s http://local-led/status?color=blue
elif echo "$line" | grep -q "Do you want to run"; then
curl -s http://local-led/status?color=orange
fi
donebash
✓ When to use
- You run long multi-file refactoring tasks with Claude Code and want to look away from your screen.
- You want to build a distraction-free physical dashboard for your vibe-coding setup.
✕ When NOT to use
- You only use Claude Code for quick, single-line edits that finish in seconds.
- You prefer keeping all project notifications strictly on-screen and digital.
What to do today
- Experiment with wrapping your IDE or terminal CLI tool inside a custom script to detect active states.
- Set up a local notification or webhook trigger when long agent tasks complete.
Sources