Widget Reference

This page provides detailed information about each widget, including its data source, what it displays, and example output.

Core Widgets

model

Property	Value
Widget ID	`model`
Data Source	stdin (model info, live `effort.level` / `fast_mode`) + settings.json fallback
Description	Displays the current model name and version with emoji. The version is read from `display_name` (`Opus 5 (1M context)` → `Opus 5`); names without one show the family only. Shows effort level for Opus/Sonnet/Fable (MAX/X/H/M/L) and fast mode indicator for Opus (↯). The live `effort.level` / `fast_mode` fields Claude Code sends on stdin win, so mid-session `/effort` changes and session-only picks are reflected; older Claude Code versions without them fall back to `settings.json`.

Example output:

◆ Opus 5.5(X)
◆ Sonnet 5(M)
◆ Opus 5.5(X) ↯
◆ Haiku 4.5

context

Property	Value
Widget ID	`context`
Data Source	stdin (context_window)
Description	Shows a progress bar, percentage, and token count for context window usage. Color changes by utilization: green 0-50%, yellow 51-80%, red 81-100%.

Example output:

-------- 45% 90K
████---- 80% 160K

contextBar / contextPercentage / contextUsage

Sub-widgets of context that render just one of its three components. Use them when you need to keep the status line compact (e.g., on a split terminal) while still surfacing the most important context information. All sub-widgets share the same data source as context, so colors and percentages stay in sync.

Widget ID	Shows
`contextBar`	Progress bar only
`contextPercentage`	Percentage only (e.g. `45%`)
`contextUsage`	Token count only (e.g. `90K/200K`)

Example layout:

"lines": [
  ["projectInfo", "contextBar", "contextPercentage", "rateLimit5h"]
]

cost

Property	Value
Widget ID	`cost`
Data Source	stdin (cost)
Description	Shows the total session cost in USD.

Example output:

$1.25
$0.03

projectInfo

Property	Value
Widget ID	`projectInfo`
Data Source	stdin (workspace) + git
Description	Shows the current directory name, git branch (clickable OSC8 hyperlink to GitHub when remote is configured), and commits ahead/behind upstream. When CWD differs from the project root, shows the relative subpath. In worktree sessions, shows a worktree indicator.

Example output:

📁 my-project (main)
📁 my-project (feature ↑3)
📁 my-project (main ↑2↓1)
📁 my-project (src/components) (main)
📁 my-project (main) 🌳 wt:feature-branch

Rate Limit Widgets

rateLimit5h

Property	Value
Widget ID	`rateLimit5h`
Data Source	API (oauth/usage)
Description	Shows the 5-hour rate limit utilization percentage with reset countdown timer. Available on both Pro and Max plans.

Example output:

5h: 42%
5h: 85% (1h23m)

rateLimit7d

Property	Value
Widget ID	`rateLimit7d`
Data Source	API (oauth/usage)
Description	Shows the 7-day rate limit utilization. Available on Pro and Max plans.

Example output:

7d: 69%

rateLimit7dSonnet

Property	Value
Widget ID	`rateLimit7dSonnet`
Data Source	API (oauth/usage)
Description	Shows the 7-day Sonnet-specific rate limit utilization. Max plan only. Deprecated ~2026-06: Anthropic merged the Sonnet weekly limit into the unified all-models weekly bucket at the Sonnet 5 launch, so the usage API returns `null` and this widget stays hidden until/unless the field returns.

Example output:

7d-S: 23%

rateLimit7dFable

Property	Value
Widget ID	`rateLimit7dFable`
Data Source	API (oauth/usage)
Description	Shows the 7-day Fable-specific weekly usage cap. Max plan only. Sourced from the API’s generic `limits[]` array (a `weekly_scoped` entry) rather than a flat field, so the widget stays hidden unless the account has a Fable cap.

Example output:

7d-F: 18%

Session Widgets

sessionId

Property	Value
Widget ID	`sessionId`
Data Source	stdin (session_id)
Description	Shows a short 8-character session identifier.

Example output:

abc12345

sessionIdFull

Property	Value
Widget ID	`sessionIdFull`
Data Source	stdin (session_id)
Description	Shows the full UUID session identifier.

Example output:

abc12345-6789-0def-ghij-klmnopqrstuv

sessionDuration

Property	Value
Widget ID	`sessionDuration`
Data Source	file (session tracking)
Description	Shows how long the current session has been running.

Example output:

45m
1h23m
2h05m

sessionName

Property	Value
Widget ID	`sessionName`
Data Source	transcript (JSONL)
Description	Shows the session name set via the /rename command. Hidden when no name has been set.

Example output:

» feature-auth
» bug-fix-123

lastPrompt

Property	Value
Widget ID	`lastPrompt`
Data Source	transcript (JSONL)
Description	Shows the most recent user prompt in the current session with a timestamp. Useful for quickly identifying the session context. Hidden when no user prompt has been sent yet.

Example output:

💬 14:32 Fix the authentication bug in middleware
💬 09:15 Add unit tests for the API client

configCounts

Property	Value
Widget ID	`configCounts`
Data Source	filesystem + stdin
Description	Counts and displays the number of CLAUDE.md files, AGENTS.md files, rules, MCP servers, hooks, and added directories configured in the project.

Example output:

CLAUDE.md: 2 | AGENTS.md: 1 | rules: 3 | MCPs: 1 | hooks: 0 | +Dirs: 2

Activity Widgets

toolActivity

Property	Value
Widget ID	`toolActivity`
Data Source	transcript (JSONL)
Description	Shows the number of running and completed tool calls in the current session. When tools are running, displays the tool name with its target (e.g., file path or command).

Example output:

⚙️ 12 done
⚙️ Read(app.ts) (8 done)
⚙️ Read(app.ts), Bash(npm test) (12 done)

agentStatus

Property	Value
Widget ID	`agentStatus`
Data Source	transcript (JSONL) + environment
Description	Shows the running subagent (type, task description) and the number of completed subagents. When the subagent’s model can be determined — from the Agent tool’s per-invocation `model` parameter, or from `CLAUDE_CODE_SUBAGENT_MODEL` for the built-in `general-purpose`/`claude` types (with `CLAUDE_CODE_SUBAGENT_MODEL_FORCE=1`, every type except `fork` and `Explore`) — it appears as a suffix, e.g. `Explore(Sonnet)`. No suffix means the subagent inherits the main conversation’s model or the model isn’t knowable from the transcript.

Example output:

🤖 Agent: Explore(Sonnet): Searching codebase +1
🤖 Agent: general-purpose(Opus): Research auth flow
Agent: 3 done

todoProgress

Property	Value
Widget ID	`todoProgress`
Data Source	transcript (JSONL)
Description	Shows the completion rate of todo items tracked in the session.

Example output:

3/5
5/5

Analytics Widgets

burnRate

Property	Value
Widget ID	`burnRate`
Data Source	transcript (per-request usage + first entry time)
Description	Session-average token consumption per minute: input + cache write + output tokens of every main-conversation request, summed from the transcript, over the time since the transcript’s first entry (idle time and resumed sessions included, so the figure stays consistent however late the widget is enabled). Cache reads are excluded — they re-read the whole prompt prefix on every call, so counting them tracks context size rather than spend. Subagent requests are not included. Shows `0/min` at session start.

Example output:

🔥 5K/min
🔥 12K/min

tokenSpeed

Property	Value
Widget ID	`tokenSpeed`
Data Source	transcript (output tokens + turn timing)
Description	Session-average output speed: the main conversation’s output tokens over the summed wall-clock span of its requests (each measured from the triggering prompt or tool result, so time-to-first-token is included). Subagent calls are excluded. Moves slowly in a long session — use `tokenSpeedLast` for the latest response. Hidden until a request has been measured.

Example output:

⚡ 67 tok/s
⚡ 120 tok/s

tokenSpeedLast

Property	Value
Widget ID	`tokenSpeedLast`
Data Source	transcript (output tokens + turn timing)
Description	Output speed of the most recent response alone, measured the same way as `tokenSpeed`. Labelled so it can sit next to `tokenSpeed`. Short responses (e.g. a lone tool call) read low because time-to-first-token dominates their span — that is the real end-to-end rate of that request, not a parsing error. Hidden until a response has both output and a measurable span.

Example output:

⚡ last 150 tok/s

cacheHit

Property	Value
Widget ID	`cacheHit`
Data Source	stdin (context_window.current_usage)
Description	Shows the percentage of input tokens served from cache on the most recent API request. Higher values indicate better cache utilization. For the session-wide view, see `promptCache`.

Example output:

85%
42%

promptCache

Property	Value
Widget ID	`promptCache`
Preset char	`c`
Data Source	stdin (`prompt_cache`, Claude Code ≥ 2.1.251)
Description	Session-wide prompt cache health for the main conversation — the same numbers as the `Prompt cache (main)` line in `/cost`. ♨️ means the cached prefix is still within its TTL (warm), ❄️ means it has gone cold and the next request re-caches the conversation. While warm it also shows the time left before the cache goes cold (from `expires_at`; seconds in the last minute). The percentage is `hit_ratio` (cache reads over all input tokens this session), and `miss N` counts requests that missed the cache. Claude Code re-renders the status line by itself when `expires_at` is reached, so the switch to ❄️ is automatic. The minute countdown in between only moves on re-renders, which pause while the session is idle — add `"refreshInterval": 60` to your `statusLine` setting to keep it ticking. Hidden until the first API response and when the provider or gateway reports no cache tokens (`caching_observed: false`). Subagent requests are not counted.

Example output:

♨️ 4m 91% miss 2
♨️ 40s 91%
❄️ 42% miss 7

promptCacheState / promptCacheHit / promptCacheMisses

Sub-widgets of promptCache that render just one of its parts, sharing its data source. A sub-widget with nothing to show (no ratio yet, zero misses) renders empty and is dropped from the line.

Widget ID	Preset char	Shows
`promptCacheState`	`w`	Warm/cold icon + time left (e.g. `♨️ 4m`, `❄️`)
`promptCacheHit`	`h`	Session hit ratio only (e.g. `91%`)
`promptCacheMisses`	`x`	Miss count only (e.g. `miss 2`)

depletionTime

Property	Value
Widget ID	`depletionTime`
Data Source	API (rate limits) + session duration
Description	Estimates the time remaining until a rate limit is reached, based on the current session consumption rate. The calculation assumes all current utilization came from this session, so accuracy improves as the session runs longer.

Example output:

~2h (5h)
~45m (7d)

Multi-CLI Widgets

codexUsage

Property	Value
Widget ID	`codexUsage`
Data Source	Codex API (ChatGPT backend)
Description	Shows OpenAI Codex CLI usage including model name and rate limit percentages. Each window is labeled by its actual duration, which varies by plan — Plus returns 5h + 7d, Pro a single 7d. Auto-hides if `~/.codex/auth.json` is not found.

Example output:

codex o4-mini 5h:30% 7d:45%

geminiUsage

Property	Value
Widget ID	`geminiUsage`
Data Source	Gemini API (Code Assist)
Description	Shows Google Gemini CLI usage for the current model only. Auto-hides if `~/.gemini/oauth_creds.json` is not found. (Gemini CLI personal tiers retired 2026-06-18; enterprise still supported)

Example output:

gemini 2.5-pro 60%

geminiUsageAll

Property	Value
Widget ID	`geminiUsageAll`
Data Source	Gemini API (Code Assist)
Description	Shows Google Gemini CLI usage across all model buckets. Auto-hides if `~/.gemini/oauth_creds.json` is not found.

Example output:

gemini pro:60% flash:20%

antigravityUsage

Property	Value
Widget ID	`antigravityUsage`
Data Source	Antigravity API (Code Assist)
Description	Shows Google Antigravity CLI weekly quota grouped by model family, such as Gemini and Claude+GPT. Auto-hides if `~/.gemini/antigravity-cli/antigravity-oauth-token` is not found.

Example output:

🪐️ Gemini 27% (6d22h) │ Claude+GPT 39% (6d19h)

antigravityUsageAll

Property	Value
Widget ID	`antigravityUsageAll`
Data Source	Antigravity API (Code Assist)
Description	Shows Google Antigravity CLI weekly quota for every model bucket, using each model’s display name. Auto-hides if `~/.gemini/antigravity-cli/antigravity-oauth-token` is not found.

Example output:

🪐️ Gemini 3.6 Flash: 27% │ Claude Opus: 39%

zaiUsage

Property	Value
Widget ID	`zaiUsage`
Data Source	z.ai API
Description	Shows z.ai/ZHIPU GLM usage including 5-hour token usage and monthly MCP usage. Auto-hides if not detected via `ANTHROPIC_BASE_URL`.

Example output:

GLM 5h:42% MCP:15%

Insight Widgets

tokenBreakdown

Property	Value
Widget ID	`tokenBreakdown`
Data Source	stdin (context_window.current_usage)
Description	Shows a breakdown of token usage by type: input tokens, output tokens, cache write tokens, and cache read tokens.

Example output:

In 30K · Out 8K · CW 5K · CR 20K

performance

Property	Value
Widget ID	`performance`
Data Source	stdin (tokens) + session duration
Description	Shows a composite efficiency score (0-100) based on cache hit rate and output token ratio. Higher scores indicate better efficiency.

Example output:

72%
95%

forecast

Property	Value
Widget ID	`forecast`
Data Source	stdin (cost) + session duration
Description	Estimates the hourly cost based on the current session spending rate.

Example output:

~$8/h
~$2/h

budget

Property	Value
Widget ID	`budget`
Data Source	stdin (cost) + file (budget config)
Description	Shows daily spending versus the configured budget limit. Requires `"dailyBudget"` in the config file. Shows a warning at 80% and critical alert at 95%.

Example output:

$5/$15
$14/$15 !!

todayCost

Property	Value
Widget ID	`todayCost`
Data Source	stdin (cost) + file (daily cost tracking)
Description	Shows the total spending across all sessions today. Aggregates costs from the current session and any previous sessions on the same calendar day.

Example output:

💰 Today: $4.83
💰 Today: $12.50

Info Widgets

version

Property	Value
Widget ID	`version`
Data Source	stdin (version)
Description	Displays the Claude Code version. Hidden when stdin does not provide version info.

Example output:

v1.0.80

linesChanged

Property	Value
Widget ID	`linesChanged`
Data Source	git (`git diff HEAD --shortstat` + `git ls-files --others`)
Description	Shows uncommitted lines added and removed, including untracked (new) files. Resets naturally on commit. Hidden when no changes exist.

Example output:

+156 -23
+42
-15

outputStyle

Property	Value
Widget ID	`outputStyle`
Data Source	stdin (output_style)
Description	Shows the current output style name. Hidden when the output style is “default” or not set.

Example output:

concise
verbose

vimMode

Property	Value
Widget ID	`vimMode`
Data Source	stdin (vim.mode)
Description	Shows the current vim mode (NORMAL or INSERT). Auto-hides when vim mode is not enabled.
Preset Char	`m`

Example output:

NORMAL
INSERT

apiDuration

Property	Value
Widget ID	`apiDuration`
Data Source	stdin (cost.total_duration_ms, cost.total_api_duration_ms)
Description	Shows what percentage of session time was spent waiting for API responses. Useful for identifying whether a session is API-bound or tool-execution-bound.
Preset Char	`a`

Example output:

API 45%
API 72%

Status Widgets

peakHours

Property	Value
Widget ID	`peakHours`
Data Source	system clock (Pacific Time via `Intl.DateTimeFormat`)
Description	Shows whether the current time falls within Anthropic API peak hours (weekdays 5:00-10:59 AM PT). Displays a countdown to the next transition. Based on PeakClaude.
Preset Char	`p`

Example output:

Peak (3h17m)
Off-Peak (23h9m)
Off-Peak (2d17h)

tagStatus

Property	Value
Widget ID	`tagStatus`
Data Source	git (`describe` + `rev-list --count`)
Description	Shows the number of commits ahead of each matched git tag. Configure patterns via `"tagPatterns"` (default `["v*"]`). Widget hides when no pattern matches a reachable tag.
Preset Char	`t`

Example output:

v1.2.3 +5
v1.2.3 +5, beta-3 +2

slashCommand

Property	Value
Widget ID	`slashCommand`
Data Source	transcript (`<command-name>` tag in user text)
Description	Displays the slash command that started the current turn (e.g. `/superpowers:brainstorming`). Cleared when the user sends a plain-text message. Survives Claude’s tool loop — `tool_result`-only user entries do not clear the command.
Preset Char	`/`

Example output:

🎯 /superpowers:brainstorming
🎯 /claude-dashboard:setup

agentMode

Property	Value
Widget ID	`agentMode`
Data Source	stdin (`agent.name`, `agent_type`)
Description	Shows this session’s agent identity — a custom agent activated via `/agent <name>` (👤) and/or the subagent type when this session was dispatched as a subagent (🤖). Distinct from `agentStatus`, which tracks subagents spawned BY this session. Hidden when neither field is present.
Preset Char	`g`

Example output:

👤 my-coder
🤖 code-explorer
👤 my-coder · 🤖 code-explorer