dsh-plugin-glm-balance
GLM (bigmodel.cn) quota balance indicator for the DeepSeek Harness web GUI.
A quota widget beside the composer's context meter — shown only when the
session's current model provider matches the configured provider id
(default zai-coding-cn).
Features
- Two mini progress bars (hourly / weekly credit limits) with percentage and
next-reset time, styled with the harness
--dsw-*theme tokens (works with light/dark and third-party themes such as Dracula). - Polls on a configurable interval (default 30 minutes); click to refresh immediately, with a spinner while the request is in flight.
- Right-click menu: open the usage page on bigmodel.cn, or refresh.
- Auth-token states:
no token/auth invalid(click shows how to fix) / upstream failure. - Its own "GLM Balance" settings section (Settings → GLM Balance) to paste the
authorizationheader, configure the provider id, and set the auto-refresh interval.
Install
dsh plugin --profile web add dsh-plugin-glm-balance
Or straight from GitHub (no npm publish needed):
dsh plugin --profile web add https://github.com/HoodyHuo/dsh-plugin-glm-balance.git
Or install from source:
dsh plugin --profile web add E:\path\to\dsh-plugin-glm-balance
Restart dsh web afterwards (the host half registers proxy routes).
Update
git -C <repo-or-clone> pull # when installed from a clone
dsh plugin --profile web add https://github.com/HoodyHuo/dsh-plugin-glm-balance.git # re-install = update
Then restart dsh web. The token file ~/.dsh/glm-balance.json is never touched
by an update. See CHANGELOG.md for what changed in each version.
Configuration
The plugin proxies GET https://bigmodel.cn/api/monitor/usage/quota/limit
through two same-origin host routes (avoiding browser CORS):
GET/POST /glm-balance/api/config— stored credentialsGET /glm-balance/api/usage— proxied quota query
The token persists in glm-balance.json under the harness home
(~/.dsh/glm-balance.json by default):
{
"authorization": "your bigmodel.cn authorization header value",
"providerId": "zai-coding-cn",
"refreshMinutes": 30
}
Edit the file, or use Settings → GLM Balance. No restart needed for token changes; the indicator picks them up on the next poll (or click).
Notes
- The
authorizationvalue is whatever bigmodel.cn expects in the header — copy it from your browser's devtools request headers. - Quota
unitmapping follows the observed API:3→ hourly,6→ weekly. - The token is stored in plain text inside the harness home, the same trust domain as the rest of the harness configuration.
License
MIT
DeepSeek Harness Web GUI 的 GLM(智谱 bigmodel.cn)套餐余额指示器插件。
在输入框上下文用量圆环旁显示额度条,仅当会话当前模型供应商匹配配置值
(默认 zai-coding-cn)时显示。
功能
- 两条迷你进度条(5 小时 / 每周额度),含百分比与下次重置时间;颜色走
--dsw-*主题 token,自动适配明暗与第三方主题(如 Dracula); - 可配置的自动刷新间隔(默认 30 分钟),点击立即刷新并显示加载动画;
- 右键菜单:打开 bigmodel.cn 用量页面、刷新(点击其他位置关闭);
- 认证状态提示:未配置 / 认证失效(点击提示修复方法)/ 获取失败;
- 设置页独立分区「GLM 余额」:填写 authorization、供应商 ID 与自动刷新间隔。
安装
dsh plugin --profile web add dsh-plugin-glm-balance
或直接从 GitHub 安装(无需发布到 npm):
dsh plugin --profile web add https://github.com/HoodyHuo/dsh-plugin-glm-balance.git
或从源码安装(安装后需重启 dsh web,宿主半要注册代理路由)。
更新
git -C <克隆目录> pull # 以克隆方式使用时
dsh plugin --profile web add https://github.com/HoodyHuo/dsh-plugin-glm-balance.git # 重新安装即更新
然后重启 dsh web。更新不会触碰 token 文件 ~/.dsh/glm-balance.json。
各版本变更见 CHANGELOG.md。
配置
Token 存放在 harness 主目录下的 glm-balance.json(默认 ~/.dsh/glm-balance.json):
{
"authorization": "bigmodel.cn 请求头里的 authorization 值",
"providerId": "zai-coding-cn",
"refreshMinutes": 30
}
可直接编辑该文件,或在 设置 → GLM 余额 分区填写;改完无需重启,下次轮询(或点击刷新)即生效。
说明
authorization的值从浏览器 devtools 的请求头里复制;- 额度
unit映射按实测 API:3→ 小时,6→ 每周; - token 以明文存于 harness 主目录,与 harness 其他配置同一信任域。
License
MIT
No comments yet. Be the first to write one.