Bands

Watches kane-cli (KaneAI) runs in Claude Code: a band above the prompt with the kane mascot, a pane for runs, steps, failures and assurance coverage, and a nudge to test what Claude just changed

Preview images pasted into the prompt box, above the prompt, before you send them, and annotate them on macOS.

A usage panel under the prompt (model, project, context, 5-hour and weekly quota with pace warnings) with a settings toolbar, English/Chinese UI and one-click handoff prompts with a handoff history and clear & continue; a subagent board showing what each subagent and workflow agent is doing; plus a moody pixel crab above the prompt that acts out each tool and walks with your subagents...

Pixel-art pomodoro timer in a side pane: /focus [minutes]

Bandeau au-dessus du prompt : modèle, projet, état git, coût, contexte, limites, boucles et tâches en cours, avec Pixel, une mascotte qui réagit, et un panneau de réglages en direct

Zeigt über dem Eingabefeld, wie lange der Prompt-Cache noch warm ist, mit Kontext, Limits und Kosten, und hält ihn auf Wunsch in Pausen warm.

A live usage panel for Claude Code with a little pet: limit bars, pace warning, daily goal, context, tokens, cost per project, history and CSV export. English and Portuguese.

A lunchbox that fills as context grows, with /doggybag to pack a handoff for a fresh session.

Workbench side pane: greg watches your PRs (the pass), birdy counts down to your next meeting.

A band above the prompt: a pace bar and a forecast for the 5-hour and weekly plan limits.

Live pane and status-line count of the agents and subagents running in this session

Click instead of remembering: a panel of each project's own commands and playbooks, what needs you now, a wiki desk for reading proposals and findings, and where you left off. It never changes how your agents work.

/tower: one pane showing every Claude Code session on this machine — which is working, which waits for you, which is done

Claude records each assumption it makes; review them in a side pane, mark them right or wrong, and the verdicts go back to Claude.

A panel above the prompt: live work progress, the session's directories and git status, usage, context and cache bars, and switchable themes.

5-hour and weekly plan usage meters with pace markers, reset countdowns and session cost, above the prompt

A pane showing the main loop and every subagent of the session as pixel crabs, each under its role's hat: a short name, role, model, plan progress, context fill, tokens, estimated cost and time, with totals on top and the finished ones kept below; while agents work, one line above the prompt says how the run stands.

Fold this session's hook messages into one line above the prompt; open them in a scrollable pane.

Lists your prompts in a pane and jumps the transcript to a selected prompt.

Plan limits, time-to-limit pace, session cost, tokens, cache hit rate, context and model in the band above the prompt

Claude FM player: a pane drawn like the stream with the current track, /lofi, volume, a mini video window; audio via yt-dlp + ffplay

Read-only view of every user and assistant message in the current session, with collapsible tool-call details and paged loading. Open with /history.

AbovePrompt band (auto-shows the edit count) + Pane (file-grouped list with [Revert]). Click the band's [View] (or /file-history) to open the Pane and revert.

rook by TestMu AI inside Claude Code: Claude runs rook against the agent it is building and gets failures back grouped into root-cause clusters with rook's remedies, writes scenarios, a live pane with per-scenario lanes and openable failures, re-test prompts after agent edits, a status line score, /rook commands and a production-target guard.