by AppGram
AgentNotch is a sleek macOS menu bar app that lives in your Mac's notch, providing real-time visibility into your AI coding assistants. Watch as Claude Code and OpenAI Codex think, read files, and execute tools — all without leaving your editor.
# Add to your Claude Code skills
git clone https://github.com/AppGram/agentnotchAgentNotch is a sleek macOS menu bar app that lives in your Mac's notch, providing real-time visibility into your AI coding assistants. Watch as Claude Code and OpenAI Codex think, read files, and execute tools — all without leaving your editor.
See every tool call as it happens — file reads, code edits, shell commands, and more. Know exactly what your AI assistant is doing at any moment.
Track token usage (input/output) and estimated costs in real-time. Never be surprised by API bills again.
Visual distinction lets you know which AI is active at a glance.
Get notified when your AI assistant finishes a task. No more wondering "is it still thinking?"
brew tap AppGram/tap
brew install --cask agentnotch
AgentNotch-1.0.0.zip from ReleasesAgentNotch.app to /ApplicationsAdd to your Claude Code configuration to send telemetry:
# Set OTEL endpoint to AgentNotch (default port 4318)
export OTEL_EXPORTER_OTLP_ENDPOINT="http://localhost:4318"
AgentNotch listens for OTLP/HTTP on port 4318 by default, and currently decodes OTLP logs (/v1/logs) and metrics (/v1/metrics).
Codex CLI (v0.79+) uses ~/.codex/config.toml with an [otel] section (not [telemetry]). Add:
[analytics]
enabled = true
[otel]
# AgentNotch currently does not decode OTLP traces, so disable trace export to avoid noisy errors.
trace_exporter = "none"
[otel.exporter.otlp-http]
endpoint = "http://localhost:4318/v1/logs"
protocol = "binary"