summary
z.ai provider data sources: API token in config/env and quota API response parsing.
read_when
Debugging z.ai token storage or quota parsing
Updating z.ai API endpoints
z.ai is API-token based. No browser cookies.
Token sources (fallback order)
Config token (~/.codexbar/config.json → providers[].apiKey).
Environment variable Z_AI_API_KEY.
GET https://api.z.ai/api/monitor/usage/quota/limit
BigModel (China mainland) host: https://open.bigmodel.cn
Override host via Providers → z.ai → API region or Z_AI_API_HOST=open.bigmodel.cn.
Override the full quota URL (e.g. coding plan endpoint) via Z_AI_QUOTA_URL=https://open.bigmodel.cn/api/coding/paas/v4.
Headers:
authorization: Bearer <token>
accept: application/json
Response fields:
data.limits[] → each limit entry.
data.planName (or plan, plan_type, packageName) → plan label.
Limit types:
TOKENS_LIMIT → primary (tokens window).
TIME_LIMIT → secondary (MCP/time window) if tokens also present.
Window duration:
Unit + number → minutes/hours/days.
Reset:
nextResetTime (epoch ms) → date.
Usage details:
usageDetails[] per model (MCP usage list).
Sources/CodexBarCore/Providers/Zai/ZaiUsageStats.swift
Sources/CodexBarCore/Providers/Zai/ZaiSettingsReader.swift
Sources/CodexBar/ZaiTokenStore.swift (legacy migration helper)