dsh-tokenlens 1.6.6 → 1.6.7

package/CHANGELOG.md

@@ -4,9 +4,14 @@
 
 **版本记录约定**：同一发布日只保留一个章节，标题取当日**最后一个**已发布版本号；当日多次发布的改动合并到该章节，并在标题下注明当日实际发布过的版本号（npm 上仍可安装这些独立版本）。若当日已有发布，后续小改动不再单独增加版本号，而是并入下一篇日志。这样日志按天规整、不出现同日多个小版本。
 
-## 1.6.6 — 2026-11-05
+## 1.6.7 — 2026-11-05
 
-（当日已发布版本：v1.6.5 / v1.6.6）
+（当日已发布版本：v1.6.5 / v1.6.6 / v1.6.7）
+
+- **fix(commandcode): 用实时窗口用量校正滞后的 `usage summary`**——`/alpha/usage/summary` 是聚合接口，比滚动窗口**滞后几美分**（实测 summary `$1.1248` vs 实时 `$1.2802`），直接拿它反推会让额度偏小、百分比偏低约 0.2 个百分点。现在取「summary 与实时窗口用量」中的**最大值**作为本期消费下界——summary 提供锚点，实时窗口只会把估计值推向真值。实测修正后月额度精确回到 **$70.00**、百分比 `1.85%`，与该平台控制台一致
+- fix(commandcode): **仅在 summary 存在时才用窗口用量校正**——没有 summary 就没有锚点，此时周窗口可能并不覆盖整个账期，用它反推会把额度放大、百分比压向 0，因此该情况仍保持 `null`
+
+### v1.6.6 同日内容
 
 - **feat(commandcode): 补上月用量百分比**——该平台付费套餐**不下发** `monthlyCreditsGranted`（额度分母），此前月窗口只能是 `unknown`。现在同时读 `GET /alpha/usage/summary`（控制台自己用的那个接口），用两个字段反推额度：`grant = credits.monthlyCredits（余额）+ totalMonthlyCredits（本期消费）`。实测 `69.1272003646 + 0.8727996354 = 70.0000000000`（多时点复核均精确 70），与控制台前端内置的 `totalCredits: 70` 一致
 - feat(commandcode): 月窗口因此同时具备**已用 / 额度 / 百分比 / 重置时间**（`$0.99 / $70.00 · 1.4%`），与 5 小时、每周两个窗口口径一致

package/README.en.md

@@ -144,7 +144,7 @@ Host bridge route: `/api/tokenlens`; agent tool: `tokenlens_usage`.
 
 ## Development
 
-Node.js ≥ 24, zero runtime dependencies: `node --test` runs the 355 tests.
+Node.js ≥ 24, zero runtime dependencies: `node --test` runs the 357 tests.
 
 Release: `node scripts/bump-version.mjs <version> "<note>"` updates `package.json`, `PLUGIN_VERSION` and `CHANGELOG.md`; see [CONTRIBUTING.md](CONTRIBUTING.md) and [SECURITY.md](SECURITY.md).
 

package/README.md

@@ -445,7 +445,7 @@ Host 半注册 `tokenlens_usage`，参数 `action`（默认 `snapshot`）与可
 
 ## 验证
 
-- `node --test`：**355 个自动化用例**，覆盖 Host 数据层、供应商适配器（含小米 MiMo 控制台 Cookie 适配器、Command Code 月度额度反推）、统计折叠、各 harness 读取、Design Plan 周期账本与额度计价、窗口锚点切片、工作区自动识别、插件集成与浏览器端渲染 / 交互。
+- `node --test`：**357 个自动化用例**，覆盖 Host 数据层、供应商适配器（含小米 MiMo 控制台 Cookie 适配器、Command Code 月度额度反推）、统计折叠、各 harness 读取、Design Plan 周期账本与额度计价、窗口锚点切片、工作区自动识别、插件集成与浏览器端渲染 / 交互。
 - 在真实环境中验证过 **5 个 harness** 的数据源；本机统计与独立的 SQLite 查询、独立的日志解析逐项对账（差额为 0），跨 harness 合并与供应商映射算术自洽。
 - 供应商接口按各官方文档与参考客户端行为验证；未配置凭据时 chip 静默隐藏，接口异常时按可重试错误降级。
 

package/index.js

@@ -84,7 +84,7 @@ export const inject = []
 export const PLUGIN_NAME = 'dsh-tokenlens'
 
 /** Plugin version; mirrors package.json (kept here so diagnostics need no fs read). */
-export const PLUGIN_VERSION = '1.6.6'
+export const PLUGIN_VERSION = '1.6.7'
 
 /** Documented defaults; every field of the raw row config is optional. */
 const DEFAULTS = {

package/lib/providers.mjs

@@ -549,14 +549,27 @@ export function parseCommandCode({ credits, subscription, whoami, usageSummary }
   // meter jump while the account did not move, so it never does that here.
   const summary = usageSummary !== null && typeof usageSummary === 'object' ? usageSummary : {}
   const periodSpend = toNumber(summary.totalMonthlyCredits) ?? toNumber(summary.totalCredits)
+  // `/alpha/usage/summary` is an aggregate and LAGS the live windows by up to a
+  // few cents (observed $1.12 vs $1.28 live), which would understate the derived
+  // grant and therefore the percentage. When the summary is present it anchors
+  // the period, and a rolling window's `used` — always a subset of the period —
+  // can then only raise the estimate towards the truth, so the largest of them is
+  // taken. WITHOUT the summary the windows are NOT usable this way: a weekly
+  // window that does not cover the whole period would inflate the grant and drive
+  // the percentage towards zero, so that case stays `null`.
+  const liveWindowSpend =
+    periodSpend === null
+      ? []
+      : [toNumber(limits.fiveHour?.used), toNumber(limits.weekly?.used)].filter((value) => value !== null)
+  const spend = periodSpend === null ? null : Math.max(periodSpend, ...liveWindowSpend)
   const reportedGrant = toNumber(credit.monthlyCreditsGranted)
   const granted =
     reportedGrant !== null && reportedGrant > 0
       ? reportedGrant
-      : remaining !== null && periodSpend !== null
-        ? remaining + periodSpend
+      : remaining !== null && spend !== null
+        ? remaining + spend
         : null
-  const monthlyUsed = granted !== null && remaining !== null ? Math.max(0, granted - remaining) : periodSpend
+  const monthlyUsed = granted !== null && remaining !== null ? Math.max(0, granted - remaining) : spend
   const monthlyPercent =
     granted !== null && granted > 0 && monthlyUsed !== null
       ? normalizePercent((monthlyUsed / granted) * 100)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-tokenlens",
-  "version": "1.6.6",
+  "version": "1.6.7",
   "type": "module",
   "description": "dsh-TokenLens (词元透镜) — provider-aware plan capacity plus local token statistics (today/7d/30d/all-time, by harness/provider/model/workspace) for DSH.",
   "license": "MIT",