Setup

Skill eltonylfgi-blip/claude-code-usage-guard/skills/setup

One-command setup for usage-guard's real plan quota. Use this whenever the user runs /usage-guard:setup, asks to enable/activate/turn on the real 5-hour and weekly quota reading, asks to wire up the status line, or when /usage-guard:usage shows no "Real plan quota" block (only the weighted fallback). Detects the installed plugin and writes the status-line shim into the user's settings.json for them, preserving any existing status line, so they don't have to edit JSON by hand.From its SKILL.md

Install

npx -y skills add eltonylfgi-blip/claude-code-usage-guard --skill setup

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

3.3 KB, as published. Nobody here has run it

usage-guard setup

Most people install usage-guard but never wire the one-time status-line shim, so the real plan quota never turns on and the guard stays silent. This skill does that wiring for them with as little friction as possible. Prefer doing the edit for the user (with a backup) over just printing instructions — the whole point is to remove the manual step.

Steps

Locate the installed status-line shim. It lives under the user's Claude plugins dir. Run this (works on macOS, Linux and Windows):

node -e "const fs=require('fs'),p=require('path'),os=require('os');const root=p.join(os.homedir(),'.claude','plugins');let hit;(function w(d){let es;try{es=fs.readdirSync(d,{withFileTypes:true})}catch(e){return}for(const x of es){if(x.name==='node_modules'||x.name==='.git')continue;const f=p.join(d,x.name);if(x.isDirectory())w(f);else if(x.name==='usage-guard-statusline.mjs')hit=f}})(root);console.log(hit||'NOT_FOUND')"

If it prints NOT_FOUND, tell the user to run /plugin install usage-guard@cc-guard first, then retry.

Open the user's settings. Path: ~/.claude/settings.json (Windows: %USERPROFILE%\.claude\settings.json). If it does not exist, create it as {}. Always back it up first (copy to settings.json.bak) before editing.
Wire the statusLine, preserving any existing one:
- If there is NO existing statusLine, set it to run the shim directly:
```
"statusLine": { "type": "command", "command": "node \"<ABSOLUTE_PATH_FROM_STEP_1>\"" }
```
- If a statusLine.command already exists, do NOT clobber it — chain it through the env var so their line still renders after the snapshot:
```
"statusLine": { "type": "command", "command": "USAGE_GUARD_STATUSLINE='<their existing command>' node \"<ABSOLUTE_PATH_FROM_STEP_1>\"" }
```
  (On Windows the env var is set in the wrapping shell — see TUTORIAL.md for the .cmd wrapper.)
Confirm in one line. Tell the user it's wired, and that after their next couple of Claude Code messages, /usage-guard:usage will start with a real Real plan quota block. Note that real quota only appears on Claude.ai Pro/Max sessions and only after the first API response; until then the weighted fallback is used.

Rules

Idempotent. If the statusLine already points at usage-guard-statusline.mjs, change nothing — just confirm it's set.
Only touch the statusLine key. Never remove or overwrite unrelated settings. Keep the backup.
When in doubt, show don't guess. If multiple installs are found or the settings look unusual, print the exact block for the user to paste instead of editing blindly.
This wires the free real-quota reading. If it still doesn't surface after setup, the user's machine may not expose rate_limits yet — that is the case the optional hands-on setup service covers.

One thing to look at

SKILL.md

usage-guard setup

Steps

Rules

Keep looking