zai-usage — GLM Coding Plan usage CLI

Quick start

export GLM_API_KEY=...        # or ZAI_API_KEY / Z_AI_API_KEY
bunx zai-usage                # quota overview

What it reads

Four undocumented Z.ai endpoints, using your GLM Coding Plan API key only. Nothing is written anywhere; the optional AI coach receives aggregate metrics only — never your prompts or code (details).

GET /api/monitor/usage/quota/limit — plan level, 5-hour token %, monthly tool-call %
GET /api/monitor/usage/model-usage — per-model tokens and calls per window
GET /api/biz/customer-package-reset/list — reset packs and your customerId (auto-discovery)
GET /api/platform-charge-zai/bill/day — the day-level billing ledger everything else is computed from

Modes

Mode	What it tells you
(default)	5-hour quota %, per-model rolling windows, reset packs
`usage`	per-model tokens/calls across Hour/24h/7d/30d/current month
`week` / `month`	behavioral review: this span vs previous, 8 tip rules, AI coach
`bill [YYYY-MM]`	month ledger: list-price spend, billed vs plan-covered, cache savings, MoM
`benefits` · `codingplan-benefits`	lifetime plan value since first billed month (`--since YYYY-MM`)
`prices`	price points learned from your own ledger
`estimate`	what a hypothetical workload would cost on list prices
`runway`	how long until the 5h window / monthly calls run out at current pace
`compare`	diff two accounts (e.g. Lite vs Pro) side by side
`export`	raw ledger as CSV or JSONL
`check`	agent gate: `--json`, exit codes, thresholds for automations

Flags and annotated output for every mode are in the full guide.

Install

# run without installing
bunx zai-usage
npx zai-usage

# or install the single file
curl -fsSL -o ~/.local/bin/zai-usage \
  https://raw.githubusercontent.com/LogicIncZo/zai-usage/main/zai-usage.ts
chmod +x ~/.local/bin/zai-usage