Reviewed mods for Claude Code

Dictionary-based mod that shows Claude Code's slash-command descriptions, /config rows, and some UI text in Korean.

Themed cards for tool rows, tables, code, shell output and Mermaid, with headings, alerts and task lists in the skin; fifteen skins you swap live with /skin

Renders GitHub Flavored Markdown alerts, task lists, strikethrough and Mermaid diagrams in the Claude Code transcript.

Pixel pals cross the band above the prompt while Claude works: Nyan Cat, Clawd, a shoot-em-up and more. Six curated scenes plus clawd-tales' Clawd story, in Korean

Clawd acts out the spinner's word above it—cooking, dancing, pacing—with a scene for every one of 189 spinner words, no model calls.

A pixel Clawd acts out every tool call above the prompt, with helper characters for each subagent—terminal-first, no model calls.

Plays YouTube Shorts in a terminal pane while Claude works and pauses when it's done. Nothing starts until you run /reels.

A stacked bar above the prompt showing what fills your context window, plus a live line for running subagents

Your context window as a stacked bar above the prompt, colored per category as /context does, with token counts and the compaction point.

A live terminal dashboard: model vitals, cost/context, an on-call architect, permission checks and subagent cards from real session events.

claude-hud 0.10.0 as a mod: model, project, git, context, usage, tools, agents, todos, alerts, budget, history, task summary, themes.

Shows context, 5-hour, and 7-day usage as gradient bars above the prompt. Desktop/VS Code only; 12 languages supported, not Korean yet.

Live GitHub PR status in Claude Code: merge readiness, CI checks, review comments and your review queue, in a collapsible band and panes.

What each turn did, in a band above the prompt: files/lines changed, commands and failures, reads, subagents; a toast when the model loops.

Shows one live line per running code review (a codex review command, or a review subagent): model, target, elapsed time, latest output.

Shows the task list's progress in a band above the prompt, read from the todo and task tools' own calls: no tool, no prompt, no tokens.

Shows Tailscale nodes' state in a band above the prompt, with a toast when a node comes up or goes down.

Always-visible 5-hour and weekly Claude usage percentages with a reset countdown above the prompt, plus new-chat and GitHub push buttons.

Type /k-mods to browse this catalog in a pane and install or remove mods with one key

Jot down your next instructions while Claude works, then drop one into the prompt with a button

Pin frequently used snippets as colored chips above the prompt: click once to copy, twice to drop one into the prompt

A pane and prompt band listing every installed plugin's commands and settings, with favourites and Claude Code's own /config rows.

Shows Spotify's current track, progress, and synced lyrics above the prompt on macOS, with transport buttons and /music commands.

A pixel-art music player pane that streams your playlist through mpv, with a character that reacts to the music.

Step through the file edits Claude made in the last turn, one diff at a time, in a pane.

Spotify-Wrapped-style recap for a Claude Code session: run /wrapped for an animated stat reveal, saved as a shareable PNG on your Desktop.

Scores 0-100 how ready your session is to /compact, and adds a template to every compaction that preserves goals, decisions and leftovers.

A live recap above the prompt (goal, doing now, waiting on you, next), with /where for a longer bullet-point summary.

Holds irreversible commands (rm -rf, force push, DB resets, volume prunes) and shows what they would change before asking

A side pane of a repository's GitHub issues as cards, with search, label filters and a button to hand one to Claude.

Colorful, themeable rendering for Claude Code replies across 15 themes: tables, code, diagrams, charts and tool rows, with copy buttons.

Nine terminal mini-games above the prompt (snake, Tetris, 2048, and more), plus a pet that grows as Claude works.

A fighting-game combo meter: every clean tool call is a hit, every error breaks the chain. D to SSS ranks, pixel-art visuals.

Space Invaders where every wave is made from the lines Claude's last Edit/Write actually added; zero tokens, played above the prompt.

A tower defense where every enemy is one of Claude's real tool calls (Bash, Edit, web/MCP, Agent); zero tokens, played above the prompt.

Shows one live line per running subagent above the prompt: time, tool count, current action. /radar lists every agent and its messages.

Shows whether this session has a Playwright browser and who holds it, and makes subagents in the same session take turns using it.

A passive 50-minute prompt-cache reminder with keep-warm, ping-once or compact choices, each shown with a live cost estimate.

Shows how much of each request's prompt the cache served vs wrote, counts down to expiry, and suggests /compact or /clear before it lapses.

A status card above the prompt: model, effort, context, 5h/week limits, cost, directory and branch, with optional GitHub CI rows.

A taxi-meter panel above the prompt: session cost, model/tier, 5-hour and weekly limits with reset time, via /meter and /receipt.

A live weather-style forecast of the context window above the prompt, with an icon, percent, and a bar chart of recent turns.

A side pane showing the main loop and every subagent as a live tree, with the context handed down and the answer handed back for each.

See the images you paste into Claude Code: thumbnails above the prompt instead of bare [Image #1] tags.

See what you paste into Claude Code: image thumbnails and long-text previews above the prompt instead of bare placeholder tags.

A rail of your session's prompts beside the transcript or above the prompt; hover to read one, click to jump to it.

Chooses the model and reasoning effort of every subagent: ask before it starts, recommend by task kind, or apply a saved default over what the agent requested.

Records tool calls like a car's black box; replay the moments before an error or denial with /blackbox. Tokens and passwords are masked.

Shows your todo list like a car navigation: progress path, next-step guidance, 'recalculating' on plan changes, an arrival announcement.

Notifies you on your Mac, out loud, or on your phone (ntfy, Telegram, Slack, Discord) when a long turn ends or Claude needs you

Hands a long session off to a fresh one with a Haiku-written brief instead of auto-compacting, with an optional viewer page for the brief.

Auto handoff to a fresh conversation when context hits a threshold, with idle cache refresh and background project notes while you're away.

Picks one skill per prompt and loads it while hiding the listing from context; uses TypeSafe's Jev with a key, else Claude's classifier.

After each turn, suggests 2 or 3 likely next prompts above the input box; pressing 1, 2, or 3 drafts one into the empty prompt, 0 dismisses.

Picks Haiku, Sonnet, or Opus for each subagent before it starts (an external routing API, or local rules), showing each pick and its cost.

Hides API keys, emails, phone numbers and Korean ID numbers on screen while you stream or share; Claude still sees the originals

Redacts API keys, JWTs, private keys and DB strings from tool results before the model reads them; blocks placeholder reuse in Bash.

Hides pasted or tool-read secrets, emails and IPs with stable placeholders, restoring real values only right before a tool runs.

A speed camera for dangerous Bash commands: force push, rm -rf, DB drops, discarding changes, prod deploys. Asks first, or always blocks.

Watches GitHub PRs from your session—merge state, review, and required checks above the prompt—with alerts when they change.

One line above the prompt for what needs you: the next meeting, PRs awaiting review, in-progress Linear issues, and recent Slack DMs.

Your Linear projects, milestones and issues in a pane; plan, execute or product buttons send a ready prompt into the session.

Shows your assigned Linear tickets in a pane; click one to load it into the session, read details, comment, or change status and priority.

Renders a real browser inside Claude Code's terminal so you/the agent can preview web pages; needs the separate terminal-browser app.

Pins the linked Vercel project's deploy queue above the prompt, one line per deploy with phase and elapsed time.

One install for a Korean Claude Code: menu, settings and status-line translation (ko-ui)

The k-mods starter set: ko-ui, skins, pixel-pals, ctx-strip and mod-store in one install

Claude Code as a taxi ride: meter (cost and limits), navigation (todo route), speed camera (risky-command check) and dashcam (tool-call recorder) in one install