feat: add heat-colored progress bars with in-bar overlay to zcode-quota

README.md

@@ -91,20 +91,39 @@ automatically. Point `ZCODE_BIN` at the bundled `zcode.cjs`:
 ## Standalone Quota CLI
 
 Besides the ACP server, the package ships a `zcode-quota` bin that queries
-your GLM Coding Plan usage **from the terminal** — no editor or running server
-needed. It reads the same `~/.zcode/v2/config.json` for credentials.
+your usage **from the terminal** — no editor or running server needed. By
+default it shows both **GLM Coding Plan** and **Opencode Go** in one card;
+pass a provider to focus on one.
+
+GLM credentials are read from `~/.zcode/v2/config.json`. Opencode Go
+credentials come from environment variables (the dashboard needs a browser
+cookie — see [Opencode Go setup](#opencode-go-setup) below).
 
 ```bash
-# One-shot: print the card and exit
+# Both providers (default): GLM + Opencode Go in one card
 zcode-quota
 
+# Focus on one provider
+zcode-quota glm            # GLM Coding Plan only
+zcode-quota go             # Opencode Go only (rolling + weekly + monthly)
+
 # Live monitor: clear the screen and refresh every 30s (default)
 zcode-quota -w
+zcode-quota go -w          # watch Opencode Go only
 
 # Refresh at a custom interval (seconds; minimum 10)
 zcode-quota --watch --interval 60
+
+# Plain monochrome bars (color is the default on a terminal)
+zcode-quota --plain
 ```
 
+By default the CLI renders heat-colored (green→yellow→red) progress bars with
+the usage numbers overlaid inside the bar, so each line stays short. Pass
+`--plain` (or `-p`) for the classic monochrome `█`/`░` layout. Color is also
+disabled automatically when stdout is piped or redirected, so captured output
+stays clean.
+
 The watch mode clears and redraws the card in place, like `top`/`htop`. Press
 `Ctrl-C` to exit. The 10s minimum exists because the quota API is cached for
 10s internally — a shorter interval would just keep returning the stale cached
@@ -116,6 +135,38 @@ When the package isn't globally installed, run the built file directly:
 node dist/bin/quota.js -w
 ```
 
+### Opencode Go setup
+
+Opencode Go has no JSON API for subscription usage — the CLI scrapes the
+authenticated dashboard at `opencode.ai/workspace/<id>/go`, so it needs your
+browser `auth` cookie. Credentials are read from two sources, **merged
+field-by-field with environment variables taking precedence** over the config
+file:
+
+- **Config file**: `~/.pi/agent/opencode-go.json` — same convention as the
+  `@beyona/pi-zai-usage` Pi extension, so if you already configured it there
+  you're done.
+  ```json
+  { "workspaceId": "wrk_your_workspace_id", "authCookie": "Fe26.2**your_cookie_value" }
+  ```
+- **Environment variables** (override the matching file field):
+  ```bash
+  export OPENCODE_GO_WORKSPACE_ID="wrk_your_workspace_id"
+  export OPENCODE_GO_AUTH_COOKIE="Fe26.2**your_cookie_value"
+  ```
+
+How to get the values:
+
+1. **Workspace ID** — open `https://opencode.ai`, navigate to your Go
+   workspace, and copy the `wrk_…` id from the URL
+   (`https://opencode.ai/workspace/<wrk_…>/go`).
+2. **Auth cookie** — open browser DevTools (F12) → Application → Cookies →
+   `opencode.ai` → copy the value of the cookie named `auth` (it starts with
+   `Fe26.2**`).
+
+Without credentials, the default dual-provider mode silently shows GLM only
+(no error). Running `zcode-quota go` without credentials prints a setup hint.
+
 ## ACP Registry
 
 This server is compatible with the [ACP Registry](https://agentclientprotocol.com/get-started/registry). It advertises a single `agent`-type auth method at `initialize` time — the GLM API key is read from `~/.zcode/v2/config.json` by the ZCode backend, so **no editor-side credentials are required**.

README.zh-CN.md

@@ -90,20 +90,35 @@ ZCode CLI 内置于桌面应用中，默认不会加到 `PATH`。用 `ZCODE_BIN`
 ## 独立配额查询 CLI（zcode-quota）
 
 除了 ACP server，本包还附带一个 `zcode-quota` 命令，可在**终端**里直接查询
-GLM Coding Plan 用量——无需编辑器，也无需 server 运行。它读取同一个
-`~/.zcode/v2/config.json` 获取凭证。
+用量——无需编辑器，也无需 server 运行。默认在一张卡片里同时显示
+**GLM Coding Plan** 和 **Opencode Go**；传入 provider 参数可只看其中一个。
+
+GLM 凭证读取自 `~/.zcode/v2/config.json`。Opencode Go 凭证来自环境变量
+（dashboard 需要浏览器 cookie——见下方 [Opencode Go 配置](#opencode-go-配置)）。
 
 ```bash
-# 一次性：打印卡片后退出
+# 双平台（默认）：GLM + Opencode Go 合并为一张卡片
 zcode-quota
 
+# 只看某一个 provider
+zcode-quota glm            # 仅 GLM Coding Plan
+zcode-quota go             # 仅 Opencode Go（rolling + weekly + monthly 三窗口）
+
 # 常驻监控：清屏并每 30s 刷新（默认）
 zcode-quota -w
+zcode-quota go -w          # 只监控 Opencode Go
 
 # 自定义刷新间隔（秒，最小 10）
 zcode-quota --watch --interval 60
+
+# 纯文本单色进度条（终端默认是彩色）
+zcode-quota --plain
 ```
 
+默认情况下 CLI 会渲染热力配色（绿→黄→红）的进度条，并把用量数字叠在条内，
+这样每行更紧凑。传 `--plain`（或 `-p`）切回经典的 `█`/`░` 单色布局。当 stdout
+被管道或重定向时，彩色也会自动关闭，保证捕获到的输出干净。
+
 watch 模式会原地清屏重绘卡片，效果类似 `top`/`htop`。按 `Ctrl-C` 退出。
 之所以设最小间隔 10s，是因为配额 API 内部有 10s 缓存——更短的间隔只会一直
 返回过期的缓存值，没有意义。
@@ -114,6 +129,28 @@ watch 模式会原地清屏重绘卡片，效果类似 `top`/`htop`。按 `Ctrl-
 node dist/bin/quota.js -w
 ```
 
+### Opencode Go 配置
+
+Opencode Go 订阅用量没有 JSON API——CLI 抓取的是登录后的 dashboard 页面
+`opencode.ai/workspace/<id>/go`，因此需要你的浏览器 `auth` cookie。设置两个
+环境变量：
+
+```bash
+export OPENCODE_GO_WORKSPACE_ID="wrk_你的工作区id"
+export OPENCODE_GO_AUTH_COOKIE="Fe26.2**你的cookie值"
+```
+
+获取方式：
+
+1. **Workspace ID**——打开 `https://opencode.ai`，进入你的 Go 工作区，从 URL
+   里复制 `wrk_…` id（`https://opencode.ai/workspace/<wrk_…>/go`）。
+2. **Auth cookie**——打开浏览器开发者工具（F12）→ Application → Cookies →
+   `opencode.ai` → 复制名为 `auth` 的 cookie 值（以 `Fe26.2**` 开头）。
+
+未设置这两个变量时，默认的双平台模式会**静默退化为只显示 GLM**（不报错）。
+若明确运行 `zcode-quota go` 但未配置，会打印一条配置提示。把它们加到 shell
+配置文件（`~/.zshrc` / `~/.bashrc`）即可持久化。
+
 ## ACP Registry
 
 本服务端兼容 [ACP Registry](https://agentclientprotocol.com/get-started/registry)。它在 `initialize` 时声明一个 `agent` 类型的认证方法——GLM API key 由 ZCode 后端从 `~/.zcode/v2/config.json` 读取，**编辑器侧无需配置任何凭据**。

src/backend/listener.ts

@@ -95,7 +95,9 @@ export class EventStreamListener {
       const isTimeout = resp.error.message === "timeout";
       if (!isTimeout || attempt === MAX_ATTEMPTS) {
         if (isTimeout) {
-          warn(`subscribe: all ${MAX_ATTEMPTS} attempts timed out (backend unresponsive for ~${Math.round((MAX_ATTEMPTS * 5000 + 500) / 1000)}s)`);
+          warn(
+            `subscribe: all ${MAX_ATTEMPTS} attempts timed out (backend unresponsive for ~${Math.round((MAX_ATTEMPTS * 5000 + 500) / 1000)}s)`,
+          );
         }
         throw new Error(formatSubscribeError(resp));
       }
@@ -126,6 +128,15 @@ export class EventStreamListener {
     }
   }
 
+  /**
+   * True if events are queued waiting for a poll (non-destructive). Lets the
+   * turn loop check liveness without consuming an event — used by the stall
+   * reconciliation to confirm a turn is still alive before ending it.
+   */
+  hasQueuedEvents(): boolean {
+    return this.queue.length > 0;
+  }
+
   /**
    * Wait for the next event, resolving once one arrives or `timeoutMs` elapses
    * (resolves null on timeout). Events arriving with no active waiter are