Bands

A band above the prompt graphing each rate-limit window (5 hours, week, per-model week) since its last reset.

Clawd lives above your prompt and reacts to what Claude Code is doing

While Claude works, Clawd walks above the prompt: across a starry sky in the desktop app, Ghostty and kitty, stopping to look around, wave and jump, and back and forth in half blocks in other terminals.

A live node map of every host your agent reaches: packets fly from your machine to github.com, npm, PyPI, MCP servers and unknown hosts, with a log and first-contact alerts

Before/after simulator screenshots on the same step as a UI edit, side by side in a pane, with a changed-pixel percentage and pinned baselines

Kintsugi for your session: failed operations crack a lacquer tile, and the run that fixes them repairs it with animated gold seams

Your git history as a Vignelli-style subway map: branches are coloured lines, commits are stations, merges are interchanges, and your working tree is a train on the current line

A Solari split-flap departure board for the agent's real work: tasks flip letter by letter in amber as they board, depart, get delayed or cancelled

A live braille seismograph of every edit above the prompt, quake alerts for big changes, and a heat treemap of the repo

A BIOS-style boot screen when a session starts: real checks of git, toolchains, devices, ports, context and disk, typed out and flipped to OK, WARN or FAIL

Tracks which test, typecheck, lint and build results are still true after later edits, and stops commit messages that claim checks which are stale

A departure board for gh stack: live stack pane, a one-line status above the prompt, and raw force-pushes, rebases and PR base edits on stacked branches redirected to gh stack

A LiveSplit-style timer above the prompt that auto-splits on recon, first edit, tests, green, commit and PR, with personal bests and gold segments

Stops the agent repeating the same failing command or undoing its own edits, and shows a stuck meter above the prompt

A roguelike pane played by your real session: context is HP, errors spawn monsters, green tests slay them, commits open chests, PRs are floor bosses

Enforces session and daily spend caps: a live gauge above the prompt, a wrap-up nudge at 80%, and new prompts refused at the cap

Blocks edits that weaken tests to fake a green run: removed or loosened assertions, added skips, deleted cases, swallowed errors, snapshot rewrites

Stacked progress band above the prompt: tasks, subagents, rate limits and monthly cost

A German vocabulary game above the prompt while Claude works, with words from the session's topic

English coach for Mandarin speakers: Grammarly-style fixes on every prompt and 3 native phrases to learn while Claude works.

After every turn, Claude Code prints you a receipt for everything it did while you asked for one thing.

Pendant que Claude travaille, une partie multijoueur de Modern Claudefare dans un panneau de Claude Code

Context window forecast plus 5h/7d rate-limit usage with reset countdowns, drawn above the prompt.

Stops commits and pushes under the wrong git identity (the work email) before they reach GitHub, and says exactly how to fix it.