Claude Usage Checker

Check Claude Code / Claude Max usage limits. Run when user asks about usage, limits, quota, or how much Claude capacity is left.

Security Scan

The skill's requirements and instructions match its stated purpose (it launches the local Claude CLI in a PTY and reads /usage), and it doesn't request extra credentials or install code, so it's internally coherent.

Purpose & Capability

Name and description match the runtime instructions: the skill needs the local 'claude' CLI and an interactive PTY to query /usage. Requiring the claude binary is appropriate for this purpose.

Instruction Scope

Instructions are narrowly scoped to launching the local claude CLI, sending the '/usage' command, reading output, and exiting. Minor caution: because it runs an interactive PTY and reads the process output, it could capture any text the CLI emits (including any unexpected sensitive lines). The instructions themselves do not ask for unrelated files, env vars, or external endpoints.

Install Mechanism

This is an instruction-only skill with no install spec and no code to download or execute. That minimizes supply-chain risk.

Credentials

The skill declares no environment variables, credentials, or config paths. That is proportionate to its stated goal of interacting with the local 'claude' CLI.

Persistence & Privilege

always is false and the skill does not request elevated or persistent system privileges or modify other skills. Autonomous invocation is permitted (platform default) but not combined with other red flags.

Assessment

This skill appears to do exactly what it says: run your local Claude CLI in a terminal and read the /usage output. Before installing or using it, confirm you have the official claude CLI installed and are comfortable letting the agent spawn an interactive terminal on your machine — whatever the CLI prints to the terminal can be read by the skill. Manually log in to claude first (the skill notes browser login won't work headlessly). If you are concerned about accidental disclosure, run the skill in a sandboxed environment (container or VM) or review the CLI behavior so you know what output it emits.

Launches the Claude CLI interactively (PTY) and reads the /usage output to report your Claude Code / Claude Max quota.

Prerequisites

• Claude CLI must be installed (npm i -g @anthropic-ai/claude-code) and logged in
• If running claude shows "Missing API key", the user must log in manually first: open a terminal, run claude, and complete the browser login flow
• Requires an interactive PTY — the agent will launch a local process and read its output (quota info only)

Steps

1. Launch claude with PTY
2. Wait for the welcome screen (poll until it appears)
3. Send /usage + Enter
4. Read the output (poll until usage data appears)
5. Close with Escape then /exit
6. Report the results

Commands

# Launch claude with PTY
exec pty=true command="claude"

# Wait and check log
process action=poll sessionId=XXX timeout=5000

# Send /usage
process action=send-keys sessionId=XXX literal="/usage"
process action=send-keys sessionId=XXX keys=["Enter"]

# Read output
process action=poll sessionId=XXX timeout=5000

# Exit
process action=send-keys sessionId=XXX keys=["Escape"]
process action=send-keys sessionId=XXX literal="/exit"
process action=send-keys sessionId=XXX keys=["Enter"]

Notes

• If you see "Missing API key" → tell the user to log in; browser-based login won't work headlessly
• Allow a few seconds between polls — Claude CLI starts slowly
• "Current week" = weekly reset, not daily

Output Format

Report in a table:

	Usage	Resets
Current session	X% used	today at HH:MM (timezone)
Weekly (all models)	X% used	HH:MM (timezone)
Weekly (Sonnet only)	X% used	HH:MM (timezone)
Extra usage	X% used / $X of $Y spent	date (timezone)

Always show reset times. The CLI displays them as "Resets Xpm" — convert to HH:MM format.