A terminal UI dashboard, abtop, monitors AI coding agents like Claude Code and Codex CLI, offering real-time insights into token usage, context window, rate limits, and running processes. It targets developers and researchers managing multiple AI agents, providing a consolidated view to optimize performance and prevent issues like hitting rate limits or leaving orphaned processes.

How It Works

abtop functions as a terminal UI (TUI) application, drawing inspiration from htop. It utilizes local Unix tools such as ps and lsof to discover and track AI agent sessions. The core advantage of its approach is its read-only, local-file-based operation, which eliminates the need for API keys or authentication, thereby enhancing privacy and simplifying setup.

Quick Start & Requirements

• Install: Use the provided bash installer: curl --proto '=https' --tlsv1.2 -LsSf https://github.com/graykode/abtop/releases/latest/download/abtop-installer.sh | sh. Alternatively, install via Cargo: cargo install abtop. Pre-built binaries are available on the GitHub Releases page.
• Prerequisites: Requires macOS or Linux bash environment with Unix tools (ps, lsof). Windows users must use WSL. A terminal size of 120x40 or larger is recommended; minimum 80x24.
• Setup: Run abtop --setup to install the rate limit collection hook.
• Links: GitHub Releases

Highlighted Details

• Real-time TUI dashboard for Claude Code and Codex CLI sessions.
• Comprehensive metrics: token usage, context window percentage, rate limits, child processes, open ports, and Git status.
• Orphan port detection and direct session pane jumping within tmux.
• Extensive theming support with 10 built-in themes, including colorblind-friendly options.

Maintenance & Community

The README mentions @tbouquet as a contributor for theme support. No other specific community channels, active development signals, or sponsorship details are provided in the README.

Licensing & Compatibility

Licensed under the MIT License. This license permits commercial use and integration with closed-source projects. The tool operates locally without requiring API keys or authentication.

Limitations & Caveats

Native Windows support is absent; WSL is required. Codex CLI agent monitoring lacks subagent and memory status tracking. While session summaries are generated locally, they rely on an indirect API call via claude --print. File contents and prompt text are intentionally excluded from display for privacy.