Claude Code's own report
Every badge comes from Claude Code's validator, not from us.
Level up Claude Code with tested mods for safer runs, smarter spending and more fun. One line to install.
/plugin marketplace add modmyclaude/claude-code-modsClaude builds mods from plain-English prompts. Copy one, change a line, and Claude builds you a new mod.
Build me a Claude Code mod called "ant-farm" that makes subagent work visible. What it does- While a turn runs, a slim band above my prompt shows elapsed time, subagents running vs done, and the estimated spend so far.- /antfarm opens a side pane listing every subagent this session: its task, model, a progress bar, tokens used, estimated cost, elapsed time and status (running, done, failed). Running ones on top, finished ones collapsed below.- Each subagent gets a small pixel-art ant that walks while it works and stops when it is done. One ant colour per model.- Header totals: total cost, total tokens, wall time. What is new here- Budget alarm: a budgetUsd setting (default 2.00). When the session estimate crosses it, show a toast and turn the status line amber; past 2x the budget, red.- ETA: once a running subagent has made 3 model requests, estimate its time left from its pace so far ("~1m left").- Narrow terminals: if the pane cannot open (under 144 columns), show a compact list inside the band instead of nothing. How it works- Track subagents through the agent spawn and turn step events and count tokens per model request.- Estimate cost from a small built-in price table per model and mark every cost as an estimate (≈).Every badge comes from Claude Code's validator, not from us.
Behaviour tests run against the real engine, in the terminal and the desktop app.
A second builder reviews every mod before it is listed.
If a mod talks to the internet or a model, its badge says so.
MIT licensed. Every mod, prompt and test is on GitHub.