claude-usage

What This Skill Does

The claude-usage skill provides a mechanism for OpenClaw to retrieve real-time data regarding your Claude Max subscription limits. Because Claude Code's usage information is presented within an interactive terminal UI (TUI) rather than as a standard machine-readable CLI output, this skill utilizes expect automation to spawn the process, navigate the interface, and extract key metrics. It parses terminal output to inform you of current session consumption, weekly model usage, and specific Sonnet-only quotas, ensuring you stay within your plan's bounds.

Installation

To install this skill, run the following command in your terminal: clawhub install openclaw/skills/skills/chunhualiao/openclaw-claude-usage

Ensure that you have the expect package installed on your system. On macOS, this is typically pre-installed at /usr/bin/expect. Additionally, verify that the claude CLI is installed and that you are already authenticated through your terminal environment before invoking the skill.

Use Cases

Use this skill whenever you need granular visibility into your AI service consumption. It is particularly useful for power users who leverage Claude heavily for coding tasks throughout the week. If you receive rate-limit notifications or simply want to plan your workflow to avoid hitting quota caps, this skill provides the necessary transparency. It acts as a bridge between the interactive nature of the Claude Code tool and the programmatic requirements of the OpenClaw agent.

Example Prompts

1. "Check how much of my Claude weekly quota I have left."
2. "/claude_usage - tell me if I'm close to my Sonnet rate limit."
3. "Am I going to run out of my current session tokens soon?"

Tips & Limitations

• Expect Dependency: This skill relies on expect scripts. If your environment lacks this utility, the skill will gracefully fall back to checking ~/.claude/stats-cache.json. Note that the fallback method is less precise than the primary method.
• Authentication: Ensure your terminal has an active session with Claude Code. If the claude command requires manual login, the script may fail to execute correctly.
• ANSI Parsing: The skill includes logic to strip ANSI escape codes, but extreme changes in the TUI layout by future Claude Code updates might require adjustment to the regex parsing logic used in the script.