@d3ara1n/pi-provider-zhipu-coding-plan

Zhipu AI Coding Plan provider for Pi Coding Agent — auto-discovers models and reports usage quota.

Features

• Dynamic model discovery — fetches available models from Zhipu's /models endpoint at startup
• Fallback model list — ships with known GLM model metadata when API is unreachable
• Usage quota reporting — displays token usage percentage and reset countdown in the status bar
• No manual config — reads API key from ~/.pi/agent/auth.json (set via /login)

Install

pi install npm:@d3ara1n/pi-provider-zhipu-coding-plan

Also install the status bar display if you want quota visibility:

pi install npm:@d3ara1n/pi-usage-block

Setup

Use /login in Pi and select the zhipu-coding provider to store your API key.

Or manually add to ~/.pi/agent/auth.json:

{ "zhipu-coding": { "apiKey": "your-key-here" } }

Models

Supported models include GLM-4.5 through GLM-5.1 series, with vision variants (suffix v). Model metadata (context window, max tokens, compatibility flags) is maintained in KNOWN_MODELS and updated alongside dynamic discovery.

Usage Display

When paired with @d3ara1n/pi-usage-block, the status bar shows:

Zhipu Coding 🟢53% ↺3h34m

• 53% — token quota consumed (from the TOKENS_LIMIT API)
• 3h34m — time until quota resets
• 🟢/🟡/🔴 — green/yellow/red based on 70%/90% thresholds

Lite plan shows one window (5h). Pro plan may show two (5h + weekly).