Claude HUD is a Claude Code plugin that turns the terminal statusline into a live dashboard for agent work. It shows what Claude is doing while you type: context usage, the current model, project path, git branch, active tools, running subagents, todo progress, and usage or rate-limit signals. That makes it useful for developers who run long Claude Code sessions and need quick visibility without opening another terminal pane.
The plugin uses Claude Code's native statusline API. Claude Code sends session JSON through stdin, Claude HUD formats it, and the result appears below the input area. Because it uses the native statusline path, it does not require tmux, a separate watcher, or a custom terminal layout. The default view keeps the core signal compact: model and project on the first line, then context and usage bars on the second line. Optional lines can show tool activity, agent status, todos, prompt cache, session cost, memory, and duration.
Installation happens from inside Claude Code. First add the marketplace with `/plugin marketplace add jarrodwatts/claude-hud`, then install the plugin with `/plugin install claude-hud`, reload plugins, and run `/claude-hud:setup`. The setup command writes the statusLine configuration. On Windows, the project recommends a full Claude Code restart after setup. On Linux, the README calls out a known `/tmp` cross-device install issue and suggests launching Claude Code with a custom `TMPDIR` if plugin installation fails with `EXDEV`.
Claude HUD is best for builders who want more observability during AI-assisted development. It helps answer simple but important questions: Is context close to full? Which tool is running? Did the agent start a subagent? How many todos are left? Which branch am I editing? Those details reduce the need to interrupt the model or scroll through logs.
The project is open source under the MIT license. When verified on May 14, 2026, the GitHub repository had about 22.7k stars and 1k forks, and the latest GitHub release shown was v0.0.12 from April 4, 2026. Requirements listed by the project include Claude Code v1.0.80 or newer and Node.js 18+ or Bun on macOS and Linux, with Node.js 18+ required on Windows.
For day-to-day use, Claude HUD is most helpful when the session is already busy. A short prompt does not need much dashboarding. A multi-file refactor, a bug hunt with subagents, or a long implementation task does. In those cases, the statusline becomes a lightweight control panel. You can see when context pressure is rising, whether Claude is still reading files, whether a todo list is moving, and whether the agent is spending time in tools instead of responding. That makes it easier to decide when to compact context, restart a session, or ask for a checkpoint.
The tool is also useful for teams standardizing Claude Code setups. Because the display is configured through the plugin and Claude Code statusline settings, a team can choose a minimal view for most developers and a fuller view for maintainers who regularly coordinate agents. The project remains developer-first: install from GitHub, inspect the source, and tune the configuration instead of accepting a black-box dashboard.