Usage Check

ID: usagecheck Version: 0.1.0 Status: Active (Claude Code path confirmed; Codex/Grok paths heuristic — see docs/USAGE.md) Owner: Soren (Cast Iron Forge)

---

Purpose

Claude Code, Codex, and Grok each expose usage/rate-limit data through incompatible mechanisms. /usagecheck detects which one is running, dispatches to that agent's adapter, and returns a single normalized report instead of three different ad hoc scripts.

---

Trigger

/usagecheck
/usagecheck --agent codex
/usagecheck --hook   (Claude Code hook mode — drop-in for the original stdin-JSON script)

---

Input Schema

interface SkillInput {
  agent?: "claude-code" | "codex" | "grok";  // Optional. Force detection.
  stdinText?: string;                         // Optional. Raw hook JSON (Claude Code path).
  configPath?: string;                        // Optional. usagecheck.config.json override.
}

---

Output Schema

interface SkillOutput {
  status: "success" | "error";
  agent: "claude-code" | "codex" | "grok" | "unknown";
  detection: { agent: string; method: "forced" | "env" | "path" | "none"; confidence: string };
  available: boolean;
  usage: Record<string, any> | null;
  message: string;
  timestamp: string;
}

---

Error Handling

See Skill Operator Guide — Error Handling for standard error codes.

Additional errors:

Code	Message	Handler
AGENT_UNDETECTED	No agent detected and none forced	Pass --agent explicitly

---

Full Reference

For Setup, Requirements, Configuration, Testing, Troubleshooting, and Integration:

→ See Skill Operator Guide

For workflow examples, adapter internals, and why Codex/Grok are heuristic:

→ See docs/USAGE.md