Louis-CFM/coucou

A tiny friend in your Mac's notch and on your iPhone that keeps an eye on your AI coding agents: Claude Code, Codex, Cursor, Gemini CLI, Antigravity and more. Approve from the notch or your Lock Screen.

Coucou – a tiny “notch companion” for Claude Code

What it is – Coucou is a native desktop app (macOS 15+ / Windows 10‑11) that lives in the screen’s notch (or slides out of the top edge on Windows). It shows live Claude Code sessions, lets you approve permission requests, drop files, and chat with Claude without leaving your current work.

Key features

  • Live Claude Code view – see what Claude reads, edits and runs, step‑by‑step, right from the notch.
  • One‑click approvals – permission prompts from Claude Code appear on the island with Allow / Deny buttons.
  • Terminal jump – on macOS the app can open the exact terminal window that a Claude session is using.
  • Chat & file drop – a built‑in chat window and drag‑and‑drop support let you ask Claude questions about a file or send it via email.
  • Window‑context attachment – on macOS you can drag the mascot onto any window to make that window’s content part of Claude’s context.
  • Integrations – optional hooks for Stripe, n8n, GitHub, Vercel, Resend, Notion, Cal.com, each shown as a colored version of the mascot.
  • Animated mascot (Mochi) – a handcrafted squircle with eyes that follow the cursor, blinks, emotes, and plays 28 short sound effects. It hides when idle and reappears on hover.
  • Privacy‑first – no telemetry; API keys are stored only in the OS keychain (macOS) or Credential Manager (Windows).

How it works

  • macOS – a borderless NSPanel (the “island”) driven by a tiny state machine, rendered with Swift 6 / SwiftUI Canvas at 60 fps. Claude Code hooks are injected via a small script that forwards events over a Unix socket.
  • Windows – a Tauri 2 app (Rust + TypeScript) that creates a transparent always‑on‑top window. Hooks communicate through a small coucou‑hook.exe using a named pipe.
  • Both builds are self‑contained with zero third‑party UI dependencies.

Installation

  • macOS – download the latest Coucou.zip from the Releases page, move Coucou.app to /Applications, and open it (first launch requires “Open Anyway” in System Settings because the app isn’t notarized yet).
  • Windows – the installer is temporarily unavailable due to a false‑positive detection; you can build it from source using Rust, Node 20 and the provided npm run pack script.
  • Build from source – instructions are provided for both platforms (XcodeGen for macOS, Cargo + npm for Windows).

Setup – after launching, open Settings from the menu‑bar (macOS) or system‑tray (Windows) and add your Anthropic API key, enable Claude Code hooks, and optionally configure any of the integration keys. Keys are stored securely in the OS credential store.

Audience – developers who use Anthropic’s Claude Code tool and want a lightweight, always‑visible UI to monitor and interact with their AI‑driven coding sessions without switching windows.

License – MIT for the code; the mascot, icon, sounds and media are © Louis Raillé and may not be reused without permission.

If you like the little mascot, consider starring the repo.

Related

  • Project
    awslabs/aidlc-workflows

    AI‑DLC is an AWS Labs framework that adds structured, auditable workflows to AI coding assistants (Claude, Kiro, Codex, Cursor, Opencode, Copilot). It provides a harness‑neutral core, multiple workflow profiles, human approval gates, and a detailed audit trail, enabling repeatable AI‑driven software delivery across many popular AI development environments.

  • Project
    Kiln-AI/Kiln

    Kiln is an open‑source, local‑first workbench (desktop app + Python library) that lets teams build, evaluate, optimize, and ship LLM‑based products. It covers eval creation, prompt auto‑optimization, RAG pipelines, multi‑agent hierarchies, synthetic data generation, and zero‑code fine‑tuning across 190+ models, with Git‑based collaboration and offline support.

  • Project
    wenet-e2e/wenet

    A production-oriented end-to-end speech recognition toolkit that provides full-stack solutions for streaming and non-streaming speech-to-text.

  • Project
    ApodexAI/FrontierAgent

    FrontierAgent is an open‑source Python runtime that provides a terminal UI for running autonomous LLM agents. It supports a single‑agent ReAct loop or a coordinated team of parallel sub‑agents, all operating inside a sandboxed filesystem with approval‑gated file changes, interactive instruction queuing, and full session tracing. The framework is model‑agnostic (any OpenAI‑compatible endpoint), includes web‑search and shell tools, and ships an evaluation harness for many research‑oriented benchmarks. Install via `uv`, configure an API key, and launch with `frontier-agent --mode react|agent_team`. Licensed under Apache 2.0.