
claude-statusbar
Manage cs (claude-statusbar) — switch theme/style/density, override severity colors, preview combinations, run doctor, reset config, install, upgrade (cs upgrade — the only supported upgrade path), or remove the bar, toggle fast/daemon mode, show cost or prompt-cache age, or toggle the activity segments (todos, active tool, running subagents, session duration, lines changed, git ahead/behind). Use whenever the user mentions cs, claude-statusbar, status bar, status line, 状态栏, 主题, theme switching, style switching, color customization, 余量颜色, 警告颜色, severity color, /statusbar, cs preview, cs doctor, fast mode, daemon, refreshInterval, 5h/7d window, context window display, prompt cache, todos / 待办, active tool, subagents / 子agent, session duration / 时长, lines changed / 行数, git ahead-behind / 领先落后, forecast / 预测 / 还能用多久, at-risk chip, showforecast, or asks to install / upgrade / update / 升级 / configure / diagnose / customize the bottom status line in Claude Code.
- Installs
- 117
- Last 24h
- —
- Trending rank
- —
- Security
- Not audited
Install
Works with every agent the skills CLI supports.
npx skills add https://github.com/leeguooooo/claude-code-usage-bar --skill claude-statusbar- 01
Install
Run the command in your project. The CLI asks which agents to install for.
- 02
Describe the task
Ask your agent for something this skill covers. It loads claude-statusbar automatically when relevant.
- 03
Review the result
The agent follows the SKILL.md steps, scripts and references bundled with the skill.
SKILL.md
The instructions your agent reads when the skill is triggered.
claude-statusbar control skill
Use this skill any time the user wants to inspect, change, customize, or
diagnose the cs status bar. Replaces the older individual slash commands
(/statusbar, /statusbar-theme, /statusbar-style, /statusbar-preview,
/statusbar-doctor, /statusbar-reset) — they still work but this skill
covers all of them with conversational intent.
Decision tree
Match the user's intent to the right cs command. Run it via Bash, then
give a short confirmation (one line, no lecture).
| User intent | Command |
|---|---|
| Inspect current config | cs config show |
| List themes | cs themes |
| List styles | cs styles |
Switch theme to <name> | cs config set theme <name> |
Switch style to <name> | cs config set style <name> |
| Change density | cs config set density <compact|regular|cozy> |
| Show all 27 combinations | cs preview |
| Filter preview to one style/theme | cs preview --style <s> or cs preview --theme <t> |
| Diagnose problem | cs doctor |
| Return one key to its default | cs config unset <key> (the config stores only what you changed) |
| Wipe config | cs config reset |
| Install / first-time setup | cs --setup |
| Upgrade / update to latest version | cs upgrade — NEVER guess uv tool install / pip install -U / pipx upgrade: it detects the install channel that is actually running cs and picks the right one (many users don't have uv at all) |
| Turn auto-upgrade off | cs config set auto_upgrade false (default true — a detached once-a-day check upgrades every install channel, binaries included; CLAUDE_STATUSBAR_NO_UPDATE=1 also wins) |
| Enable fast mode (daemon) | cs --setup --fast |
| Disable fast mode | cs daemon stop then re-run cs --setup |
| Toggle session cost display | cs config set show_cost true|false |
| Toggle prompt-cache countdown | cs config set show_cache_age true|false |
| Toggle project + branch 2nd line | cs config set show_project_branch true|false (default true) |
Toggle the ocs line (ocs boss · claude-7d5a5d07 · ✉3 · lan win: address, unread DMs, LAN bridge; own line; auto-hidden without ocs ≥ 0.5.0, ✉ needs an ocs whose inbox accepts --session) | cs config set show_ocs true|false (default true) |
Toggle todo progress (▸ task 3/7, 3rd line) | cs config set show_todos true|false (default true) |
Toggle active-tool indicator ◐ (3rd line) | cs config set show_tools true|false |
Toggle completed-tool rollup ✓ name×N (3rd line) | cs config set show_tool_rollup true|false (default off — volume tally) |
| Toggle running-subagent bottom line(s) | cs config set show_agents true|false (default off — Claude Code shows background agents natively) |
Toggle session duration ⏱ (on identity line) | cs config set show_duration true|false |
Toggle lines added/removed +/- (on identity line) | cs config set show_lines true|false |
Toggle git ahead/behind ↑↓ (on identity line) | cs config set show_ahead_behind true|false |
Toggle the bar_shimmer twinkling starfield (experimental, classic only) | cs config set bar_shimmer true|false (default off) |
| Toggle the rate-limit forecast (→NN% projected use / ⚠eta warning) | cs config set show_forecast true|false (default on) |
Toggle the faint version + update hint at the identity-line end (· vX.Y.Z ↑new) | cs config set show_version true|false (default on) |
| Toggle the ⚙ session-mode line (effort/thinking/fast/output-style) | cs config set show_mode true|false (default on) |
| Toggle the per-effort colour gradient on the mode line | cs config set mode_gradient true|false (default on) |
| Hide weekly bar | cs config set show_weekly false |
| Set warning threshold | cs config set warning_threshold <0-100> |
| Set critical threshold | cs config set critical_threshold <0-100> |
| Auto-collapse to hairline below width | cs config set auto_compact_width <px> |
| Force / disable no-quota (API) mode | cs config set api_mode <auto|on|off> |
| Show working directory | cs config set show_cwd true|false + cs config set cwd_style <basename|full> |
| Relay balance fuel gauge | cs config set show_balance true|false + cs config set balance_bar true|false (no-quota mode only; auto-hidden when the relay has no balance API) |
| Exit-IP purity segment (also checks IPv4 and IPv6 exits separately against our own host — warns on an IPv4/IPv6 split, WARP, VPN/proxy, or an unsupported region; never contacts Anthropic, and a per-domain proxy rule can still route claude.ai differently) | cs config set show_ip_risk true|false |
| Relay fingerprint-risk watermark | cs config set show_fp_risk true|false |
Window-end projection →NN% | cs config set show_projection true|false |
| Worktree marker glyph | cs config set worktree_glyph 🌲 (default ⑂; any 1-2 cell symbol — an emoji ignores the theme color, the emoji font decides it) |
| Worktree marker color | cs config set color_worktree "#bd93f9" (empty string restores the theme's hue) |
No-quota mode (third-party relay / Bedrock / Vertex)
When Claude Code points at a third-party relay (ANTHROPIC_BASE_URL ≠
api.anthropic.com) or a cloud backend (CLAUDE_CODE_USE_BEDROCK /
CLAUDE_CODE_USE_VERTEX), the official 5h/7d quota doesn't exist. cs detects
this and switches to a no-quota layout: the two quota battery bars are
dropped and the context window is promoted to its own ctx[…] battery bar
(green→yellow→red on 70/85% used), followed by the model name + the usual
live-activity tail. This mirrors claude-hud's behavior and is what to reach for
when a user says "用 API 就没状态了 / 连上下文都没了".
- Detection is automatic (
api_mode = auto, the default). A transcript-based heuristic also catches relays whose env var didn't reach the statusLine subprocess (an assistant turn exists yet quota never arrived → no-quota). - Force it on a setup where auto-detect misses:
cs config set api_mode on(or per-shellCS_API_MODE=on). Force the official layout back withapi_mode off.CS_API_MODEenv wins over the saved config. - Works under both the inline and fast-mode (daemon) render paths.
Per-severity color overrides (v3.4.1+)
The user can override the three severity colors independently of theme:
cs config set color_ok "#4ec85b" # calm / safe
cs config set color_warn "#e8b260" # warning
cs config set color_hot "#e87474" # critical
cs config set color_ok "" # clear back to theme default
Accepts #rrggbb, #rgb, or bare rrggbb. The override layers on top of
whatever theme is active — no need to switch theme just to tune one color.
When user says "make 余量颜色 / safe color / green sharper", "warning 偏淡", "critical too red" — go to the override, not the theme.
Vibe → theme suggestion
If the user describes a vibe instead of naming a theme, suggest one and ask before switching:
| Vibe / context | Theme |
|---|---|
| Muted, professional, dark terminal | graphite (default) |
| Soft, warm, dark | twilight |
| Classic dev / Nord-inspired | nord |
| High contrast, vivid | dracula |
| Warm, cute, light bg | sakura |
| Light terminal | linen |
| Pure grayscale / no color | mono |
| Popular pastel, easy on long viewing | catppuccin-mocha |
| Deeper neon-blue mood | tokyo-night |
Render anatomy (so you can explain what you're changing)
5h[██16% ░░░]⏰2h27m | 7d[██32% ░░]⏰4d05h | Opus 4.7(280k/1M) | $ 1.42 | cache 4m23s
└─ 5h ─────┘└──5h──┘ └─ 7d ──┘ └─7d─┘ └────context───────┘ └cost┘ └─cache─┘
Every numeric segment colors itself by its own severity (since v3.4):
- 5h →
theme.s_*chosen frommsgs_pct - 7d →
theme.s_*chosen fromweekly_pct - model+context →
theme.s_*chosen fromctx_used_pct(None → neutral) - cache → its own string-age severity (COLD → red, <1m → yellow, else green)
[ ],( ),|→theme.mute(recede behind data)
Common diagnostic flows
"Status bar isn't showing" → cs doctor. It self-checks:
- Claude Code's
~/.claude/settings.jsonhas thestatusLineblock - the
csbinary is on PATH - whether the daemon is alive (if fast-mode configured)
- whether the cache files are stale
"refreshInterval too high" → cs doctor will recommend cs --setup --fast
when it sees refreshInterval ≤ 2s on the inline command. Fast mode drops
1Hz CPU from ~6% to ~2%.
"Color won't change after cs config set theme X" → check the user
isn't on a Claude Code session that read settings.json at start. Ask
them to send a new prompt; the next render picks up the new theme.
Don't
- Don't lecture about ANSI codes or terminal compatibility unless asked.
- Don't suggest editing
~/.claude/claude-statusbar.jsonby hand. Usecs config set <key> <value>. - Don't change theme just to fix one color — use
color_ok / color_warn / color_hotoverrides. - Don't run destructive commands (
cs config reset,cs daemon stop) without confirming with the user.
Style of response
Be terse. Run the command, paste the one-line confirmation, point at the next step if relevant. The status bar is on screen — they can see the result immediately, no need to describe it.
FAQ
What does the claude-statusbar skill do?
Manage cs (claude-statusbar) — switch theme/style/density, override severity colors, preview combinations, run doctor, reset config, install, upgrade (cs upgrade — the only supported upgrade path), or remove the bar, toggle fast/daemon mode, show cost or prompt-cache age, or toggle the activity segments (todos, active tool, running subagents, session duration, lines changed, git ahead/behind). Use whenever the user mentions cs, claude-statusbar, status bar, status line, 状态栏, 主题, theme switching, style switching, color customization, 余量颜色, 警告颜色, severity color, /statusbar, cs preview, cs doctor, fast mode, daemon, refreshInterval, 5h/7d window, context window display, prompt cache, todos / 待办, active tool, subagents / 子agent, session duration / 时长, lines changed / 行数, git ahead-behind / 领先落后, forecast / 预测 / 还能用多久, at-risk chip, showforecast, or asks to install / upgrade / update / 升级 / configure / diagnose / customize the bottom status line in Claude Code.
How do I install claude-statusbar?
Run npx skills add https://github.com/leeguooooo/claude-code-usage-bar --skill claude-statusbar in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/claude-statusbar/ (personal) or .claude/skills/claude-statusbar/ (project).
Which AI agents can use claude-statusbar?
Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.
Is claude-statusbar safe to install?
No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.
Who maintains claude-statusbar?
It is published by leeguooooo in leeguooooo/claude-code-usage-bar. It has 117 recorded installs.
Popular ai & agents skills
google-agents-cli-publish
google/agents-cli
This skill should be used when the user wants to "publish an agent", "publish my ADK agent", "register an agent with Gemini", "publish to Gemini" (formerly Gemini Enterprise), or needs guidance on the agents-cli publish gemini command. Also use when the user wants to "manage agents in Agent Registry", "list/update/delete registered agents", or "register an MCP server". Covers ADK vs A2A registration modes, programmatic and interactive usage, flag reference, auto-detection from deployment metadata, Agent Registry fleet management, and troubleshooting. Part of the agents-cli skills suite. Do NOT use for deployment (use google-agents-cli-deploy).
google-agents-cli-eval
google/agents-cli
This skill should be used when the user wants to "run an evaluation", "evaluate my agent", "evaluate my ADK agent", "write an eval dataset", "analyze eval failures", "compare eval results", "optimize agent", or needs guidance on the Gemini platform eval methodology and the Quality Flywheel. Covers eval metrics, dataset schema, LLM-as-judge scoring, and common failure causes. Applies to any agents-cli project, whatever framework the agent is written in. Do NOT use for agent API code patterns (ADK: use google-agents-cli-adk-code), deployment (use google-agents-cli-deploy), or project scaffolding (use google-agents-cli-scaffold).
skill-creator
anthropics/skills
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
codex-pet
prime-skills/runcomfy-agent-skills
Codex Pet generator on RunComfy. Build a Codex-compatible Codex Pet spritesheet.webp + pet.json from a single reference image, drop it into ${CODEXHOME:-$HOME/.codex}/pets/ / and Codex picks it up as a custom Codex Pet next to the 8 built-ins. This skill produces the exact Codex Pet atlas Codex expects (1536x1872 PNG/WebP, 8 cols x 9 rows, 192x208 cells, 9 animation states — idle, running-right, running-left, waving, jumping, failed, waiting, running, review). Calls OpenAI GPT Image 2 edit ONCE via the local RunComfy CLI as runcomfy run openai/gpt-image-2/edit to produce a canonical Codex Pet pose, then assembles all 9 animation rows programmatically with ImageMagick micro-transforms — no Codex Pro, no $imagegen, no OPENAIAPIKEY required, only RUNCOMFYTOKEN. Triggers on "codex pet", "create codex pet", "make codex pet", "hatch codex pet", "/hatch image", "desktop pet codex", "codex pets", "spritesheet.webp", or any explicit ask to build a custom pet for OpenAI Codex.
full-output-enforcement
leonxlnx/taste-skill
Overrides default LLM truncation behavior. Enforces complete code generation, bans placeholder patterns, and handles token-limit splits cleanly. Apply to any task requiring exhaustive, unabridged output.
executing-plans
obra/superpowers
Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available