A cockpit for Claude Code: plan bars and subagent strips, context, usage limits with reset countdowns and burn rates, spend, the model route and a pet. Six bar styles, three collapsed looks. CLI and Desktop.
What this mod can do
- Reviewed at commit 5e07b94
- Runs programs
- Writes files
- Reads environment variables
- Sees each prompt you send and can change or drop it
- Sees each tool call before it runs and can block or change it
- Sees each subagent before it starts and can stop it or pick its model
- Also uses prompt.autocomplete, which this page doesn't describe yet
9 more things it can do
- Reads files
- Reads your settings
- Reads your prompt as you type
- Remembers data between sessions
- Plays sounds
- Adds the /pet command
- Lists running subagents
- Runs on a timer
- Runs on Claude Code hooks, such as session start
Good to know (3)
- Needs Claude Code 2.1.293 or later.
- Reads your prompt as you type to react to it, in memory only. It never saves or sends it.
- Saves settings in ~/.claude/pixel-pet.json, with project paths if a project gets its own pet.
Preview
Terminal
▀▀▀▀▀▀▀▀▀▀▀▀▀▀ [-]
▀▀▀▀▀▀▀▀▀▀▀▀▀▀
██▀▀▀▀██▀▀▀▀█▀ Noodle is humming quietly
▀▀▀█▀▀██▀▀█▀▀▀ ━━━─────────── 22% 5h ▰▱▱▱▱ 26% week ▰▱▱▱▱ 7% $0.20 ⓘ
▀▀▀▀█▀▀▀▀█▀▀▀▀Claude Desktop
What it does
In the author's words
An animated pixel-art companion above your prompt that reacts to your session and your typing, with a live context meter.
- What you see
- A pet and a context meter above the prompt
- Works in
- Terminal and Claude Desktop
- Slash commands
- /pet
- Uses your tokens
- No
- Changes files
- ~/.claude/pixel-pet.json, plus an SVG from /pet card
- Network access
- Only through programs it runs
- Reviewed commit
- 5e07b94
- Preview recorded with
- Claude Code 2.1.309
How it works
- Starts asleep in a new session and wakes on your first keystroke.
- Reads hook events (tool calls, prompts, turns, subagents) and Claude Code's usage figures.
- Prints one line when the kind of work changes, and cheers or groans at test results.
- Draws a context bar beside it, then your plan's 5-hour and weekly use, or API dollars.
Common questions
How can I tell when Claude Code needs my approval?
Pixel Pet jumps up waving with an exclamation mark when a permission prompt appears, and says it needs your OK. When a turn of 30 seconds or more finishes, it waves until you're back. /pet sound on adds a soft chirp when it calls you.
How do I see my context and plan usage with Pixel Pet?
The meter on its second line is a bar for the context window, then your plan's 5-hour and weekly usage, or the dollars billed on the API, Bedrock, Vertex or Foundry. Point at the info icon for tokens, reset times and how you're billed. It notes once when a plan limit passes 80%, and again at 95%.
Can I turn off the animation or the background?
/pet calm on keeps motion small, with no bobbing or scrolling. /pet scene off shows just the pet, and /pet hide tucks it away until you run /pet show.
Does Pixel Pet send anything off my machine?
No. It makes no network calls. It runs date +%z once a session for your time zone, writes ~/.claude/pixel-pet.json, and /pet card saves an SVG to your Downloads folder.
More like this
One line above the prompt: context tokens with their weather, a bar per recent prompt, your 5-hour and 7-day limits against the time elapsed, the prompt cache with its time left, the session cost with the last prompt's share in dollars and in points of the 5-hour limit, and the subagents running. Labels in English, French or Russian. After Token Weather (Anthropic, Apache-2.0) and the ideas of usage-meter (HolyGrail) and prompt-cache-control (Daniel San).
Flightdeck: a live agent dashboard for Claude Code. Main model vitals, an on-call architect, every permission check, subagent cards and swimlanes, a turn receipt and a session log, all from real session events
Live session cost odometer above the prompt: a burning fuse, real-world comparisons, threshold alerts
Opus 4.8 [high] ~/app main
16:20:55 | ⛁ ██░░░░░░░░ 22% | 5h ●◔○○○ 26% ↻2h7m | 7d ◔○○○○○○ 7% ↻2d1h │ $0.41 ⏱ 10m
Two-line status with a block context bar and quarter-step dot meters (◔◑◕●) for the 5-hour and weekly limits (fill shows usage, color shows whether you are ahead of or behind pace), plus model, effort, git and cost. Needs no jq.
Opus 4.8 high ✦ │ ctx 22% │ app (main) │ PR #1287 ● │ $0.41 · 10m +128 -34
Session ● ● ● ● ● ● ● ● ○ ○ ○ 74% left Resets in 2h 6m
Weekly ● ● ● ● ● ● ● ● ● ● ○ 93% left Resets in 2d 0h
Fable ● ● ● ● ● ● ● ● ● ○ ○ 85% left Resets in 2d 0h
◐ Read:index.ts ✓ Read×1 ✓ Bash×1 ✓ Edit×1 ✓ Grep×1
◐ code-reviewer Review the change for regressions
A header with model, context and burn rate per hour, then live tool and agent activity parsed from the session transcript.
Model: Opus 4.8 | Thinking: high | Context: [▰▰▱▱▱▱▱▱▱▱] 45k/200k (22%) | cwd: ~/app | ⎇ main | (+0,-0)
Cost: $0.41 | Block: 2hr 6m | Weekly: 7.0% | Session: 10m | In: 916.7 t/s | Out: 29.2 t/s | Cached: 2k | Session: [▰▱▱▱▱] 26.0%
Mem: 0.8G/1.0G | (*ﾟ∀ﾟ*) | Session ID: b8e1c0d2-4a6f-4e2a-9c1b-3f5d7a9e2c10
Three lines in One Dark: model, thinking effort and a context bar up top; cost, quota, session clock and token throughput in the middle; weather, free memory and a rotating kaomoji below.
app · main · Opus 4.8 · high
ctx 22% 44k · 5h 26% ↻2h6m · 7d 7% ↻2d0h · cost $0.41 · msg $0.757Two-line: dir/git/model/advisor/effort on top + Caveman, RTK, context %, 5h/7d rate limits, and session cost below