dsh-cost-meter 1.5.42 → 1.6.2

package/README.en.md

@@ -4,9 +4,9 @@
 
 **Session cost tracking plugin for the DeepSeek Harness web GUI (bilingual UI)**
 
-Per-conversation cost · daily totals · OpenCode Go subscription quota display · budget with usage percentage · official account balance · custom provider balance · balance progress bar · history · peak/off-peak pricing hours display (peak hours UTC 01:00–04:00, 06:00–10:00; from Aug 23, 2026 weekends are billed at off-peak prices all day, shown as “Weekend — all off-peak”) · pre-switch popup & system-notification alerts for peak/off-peak changes (position / lead time / alert type configurable) · one-click price sync from the official docs · Codex-style token usage heat grid · multi-vendor model pricing (built-in 90+ model price catalog with auto-matching) · mainstream Coding Plan quota queries & display (Anthropic / Z.ai / MiniMax / Kimi / OpenRouter / SiliconFlow / CommandCode / SCNet) · quota strip above the input box (budget / Go / coding-plan usage in one row, toggleable)
+Per-conversation cost · daily totals · OpenCode Go subscription quota display · budget with usage percentage · official account balance · custom provider balance · balance progress bar · history · peak/off-peak pricing hours display (peak hours UTC 01:00–04:00, 06:00–10:00; from Aug 23, 2026 weekends are billed at off-peak prices all day, shown as “Weekend — all off-peak”) · pre-switch popup & system-notification alerts for peak/off-peak changes (position / lead time / alert type configurable) · one-click price sync from the official docs · Codex-style token usage heat grid · multi-vendor model pricing (built-in 90+ model price catalog with auto-matching) · mainstream Coding Plan quota queries & display (Anthropic / Z.ai / MiniMax / Kimi / OpenRouter / SiliconFlow / CommandCode / SCNet) plan/API dual-track billing (subscription quota vs pay-as-you-go money separated, per-1% & full-window token/equivalent-cost estimates with daily/weekly/monthly curves) · · quota strip above the input box (budget / Go / coding-plan usage in one row, toggleable)
 
-[![version](https://img.shields.io/badge/version-1.5.42-4176E6)](https://github.com/Han-1413141/dsh-cost-meter)
+[![version](https://img.shields.io/badge/version-1.6.2-4176E6)](https://github.com/Han-1413141/dsh-cost-meter)
 [![npm](https://img.shields.io/npm/v/dsh-cost-meter?label=npm)](https://www.npmjs.com/package/dsh-cost-meter)
 [![license](https://img.shields.io/badge/license-MIT-green)](LICENSE)
 [![dsh](https://img.shields.io/badge/DeepSeek%20Harness-dsh--plugin-4176E6)](https://github.com/deepseek-ai/deepseek-harness)
@@ -36,6 +36,7 @@ English | [中文](README.md)
 | Budget box | Sidebar bottom (between the balance row and the settings button) | Rounded-square frame: budget, used %, progress bar, today's cost & share of budget, used/limit; ≥80% warning, ≥100% over-budget |
 | Summary cards | Settings page | Today / this month / cumulative cost and call counts |
 | Token usage stats | Settings page (Cost section) | All-time token totals (input/cache/output/calls) + a Codex-style 26-week daily usage heat grid that fills the settings width; hover a cell for that day's detail |
+| Token Plan usage stats | Settings page (Usage) | Per enabled coding plan (incl. Go): per-1% quota and full-window token / equivalent-cost estimates for the current windows (sample delta / live ratio), plus daily/weekly/monthly usage curves; plan-channel amounts are equivalent-only and never touch real money (issue #64) |
 | Today's sessions | Settings page | Per-session call count, input/cache/output tokens and cost |
 | History | Settings page | Per-day totals; retention days configurable (default 180) |
 | Pre-install history import | Automatic on first launch | After install/upgrade, the first launch automatically replays all host session logs to import conversations from before the plugin was installed (missing dates are rebuilt whole; existing dates only gain previously unknown sessions; idempotent and never double-counts live metering; costs priced at per-event historical rates); a manual re-run entry remains in Settings |
@@ -49,6 +50,7 @@ English | [中文](README.md)
 | AI price sync | [prompt](docs/AI-PRICE-SYNC-PROMPT.en.md) | DeepSeek official sync; other providers use the verified official price catalog and manual configuration |
 | Model & Plan adaptation guide | [adaptation doc](docs/model-and-plan-adaptation.en.md) | Adaptation matrix for per-model billing and the 8 Coding Plan vendors, the auto-matching mechanism and price sources ([中文](docs/model-and-plan-adaptation.md)) |
 | Peak/off-peak alert guide | [alert doc](docs/peak-alert.en.md) | Fully illustrated guide to the pre-switch popup and system notification: effect screenshots (EN/中文), settings reference and usage tips ([中文](docs/peak-alert.md)) |
+| Token Plan usage stats guide | [panel doc](docs/token-plan-stats.md) | Meaning of the four columns in the per-1% & full-window panel, the end-to-end delta estimation method and precision tags, the scope boundary (dsh-made calls only) and usage curves ([中文](docs/token-plan-stats.md#中文)) |
 | Multi-provider billing | Settings / ledger | OpenAI, Anthropic, Google Gemini, Mistral and other providers with input/output, cache and reasoning-token pricing isolated by provider + model |
 | Model-name auto-matching | Settings / ledger | Unknown model ids are matched against the price table: case/spaces/hyphens/dots and bracket annotations (e.g. (go)) are ignored — a normalized-equal or containing name hits (e.g. `gpt5.6 luna(go)`); router providers (opencode/zen etc.) search across all vendors; can be restricted to exact match, and unmatched models can be pinned to a specific entry in Settings |
 | Extended price catalog | Settings → Extended price catalog | Built-in reference catalog grouped by vendor and model family (expandable; vendors collapsed by default); mount entries into billing with one click — mounted third-party models live inside the catalog and stay editable; a per-model “Show directly in Cost settings” toggle chooses which models (DeepSeek included) appear directly in the price table |
@@ -229,22 +231,22 @@ Real captures from an actual DSH sidebar of the period strip and collapsed verti
 dsh plugin --profile web add dsh-cost-meter
 ```
 
-**PowerShell one-click script** (copy the whole line, paste, press Enter; pnpm is provisioned automatically, git is auto-detected — no clone needed; the install chain is **pinned to the release tag `v1.5.42`** — review the script before running):
+**PowerShell one-click script** (copy the whole line, paste, press Enter; pnpm is provisioned automatically, git is auto-detected — no clone needed; the install chain is **pinned to the release tag `v1.6.2`** — review the script before running):
 
 ```powershell
-irm https://raw.githubusercontent.com/Han-1413141/dsh-cost-meter/v1.5.42/install.ps1 | iex
+irm https://raw.githubusercontent.com/Han-1413141/dsh-cost-meter/v1.6.2/install.ps1 | iex
 ```
 
 **Or a plain command line** (the machine must already have pnpm and git; also pinned to the tag):
 
 ```sh
-dsh plugin --profile web add github:Han-1413141/dsh-cost-meter#v1.5.42
+dsh plugin --profile web add github:Han-1413141/dsh-cost-meter#v1.6.2
 ```
 
 Without git, use the GitHub tag archive:
 
 ```sh
-dsh plugin --profile web add https://github.com/Han-1413141/dsh-cost-meter/archive/refs/tags/v1.5.42.tar.gz
+dsh plugin --profile web add https://github.com/Han-1413141/dsh-cost-meter/archive/refs/tags/v1.6.2.tar.gz
 ```
 
 After installing, **restart** `dsh web` (plugin rows, the Typert manifest and the client bundle are all scanned at startup):
@@ -281,6 +283,9 @@ dsh plugin --profile web add link:./dsh-cost-meter  # symlink; edit lib/client.j
 - The session badge is **billed exactly** at the moment each call is made (host-exported per-call cost), just like daily/monthly/cumulative totals and the budget;
 - Billing sources are the `usage` block of every model call (including sub-agents, compression, title generation and other auxiliary calls), matching the billable view;
 - Budget and over-budget warnings **only warn — they never block calls**.
+- **Plan/API dual-track billing** (issue #64): calls on subscription-style channels (MiniMax, manually-marked Codex, etc.) are recorded as catalog-price equivalents only; budget/today-cost/overview cards count the pay-as-you-go (API) channel exclusively.
+- **Full-repo security audit fixes (v1.6.2)**: resolved 3 high-risk findings — ledger writes lost on exit (close/flush ordering), command injection in the release script, panel null-pointer crash on null Go monthly window — plus 30+ medium/low issues including missing hour-bucket entries for routed calls, dead balance-reconciliation warnings, failed custom-balance extraction silently showing $0, and leaked upstream streams on consumer abort; ledger corruption auto-backup, write-failure retry, config-patch atomicity and prototype-chain hardening included. Itemized list: [docs/security-audit-v1.6.2.md](docs/security-audit-v1.6.2.md).
+- **"Include plan totals" toggle** under the overview cards switches every amount between real-money (API channels only, default) and total equivalent cost (plan subscriptions included); routed calls with a missing provider whose models match third-party catalogs are auto-classified into the matching plan, and legacy per-day residuals without model details count toward the API scope; The Token Plan stats panel in Settings → Usage provides per-1% and full-window token/equivalent-cost estimates plus daily/weekly/monthly usage curves; classification is configurable per vendor and per provider:model.
 
 ## Data storage
 
@@ -351,6 +356,8 @@ dsh --profile web --port 3099                           # real startup (watch lo
 - Price sync overwrites the same-named models listed on the official page; custom model entries are unaffected;
 - Balance lookup needs network access to api.deepseek.com and a valid API key; **the API key is only ever sent to the official domain** (if baseURL points at a non-official host, balance queries refuse to run — model requests are unaffected);
 - The OpenCode Go quota endpoint is the official opencode.ai endpoint (community-documented); if its response shape changes, the Settings page shows an error and the display can be turned off in Display settings;
+- The per-1% quota / full-window figures in Token Plan stats are estimates: vendors only report percentages with ones-digit quantization (a displayed 1% may really be 0.5%–1.5%), so the plugin derives them from end-to-end deltas over continuous sample segments to suppress quantization noise (segments spanning <5 percentage points are tagged "low reading precision"; samples older than 7 days fall back to live ratio; window edges align to whole hours). Server percentages cover ALL usage of the account — consumption of the same key on other machines/CLIs is not in the local ledger, so low estimates are expected; treat them as cross-plan comparisons only;
+- Plan/API classification is inferred from channel and config; mixed subscription/PAYG usage of one vendor (e.g. Kimi) can be overridden per provider:model in settings.
 - A restart of `dsh web` is required after installing/updating the plugin.
 
 ## Update history

package/README.md

@@ -4,9 +4,9 @@
 
 **DeepSeek Harness 会话费用统计插件(界面中英双语)**
 
-本会话费用 · 当日费用 · OpenCode Go 订阅额度显示 · 预算与已用百分比 · 官方账户余额 · 自定义 Provider 余额查询(可配任意 HTTP 端点) · 余额三段进度条 · 历史记录 · 峰谷计价时段显示(UTC 01:00–04:00、06:00–10:00 为峰时段;2026-08-23 起周末全天按谷价,显示「周末时段——全谷价」) · 峰/谷切换前弹窗与系统通知提醒(位置/提前量/提醒类型可配) · 官方价格一键同步 · 类 Codex Token 用量热图 · 多厂商多模型价格计费(内置 90+ 模型价格目录与自动匹配) · 主流 Coding Plan 额度查询与显示(Anthropic / Z.ai / MiniMax / Kimi / OpenRouter / SiliconFlow / CommandCode / SCNet 八家) · 输入框上方额度横条(预算/Go/Coding Plan 用量一条横排显示,可开关)
+本会话费用 · 当日费用 · OpenCode Go 订阅额度显示 · 预算与已用百分比 · 官方账户余额 · 自定义 Provider 余额查询(可配任意 HTTP 端点) · 余额三段进度条 · 历史记录 · 峰谷计价时段显示(UTC 01:00–04:00、06:00–10:00 为峰时段;2026-08-23 起周末全天按谷价,显示「周末时段——全谷价」) · 峰/谷切换前弹窗与系统通知提醒(位置/提前量/提醒类型可配) · 官方价格一键同步 · 类 Codex Token 用量热图 · 多厂商多模型价格计费(内置 90+ 模型价格目录与自动匹配) · 主流 Coding Plan 额度查询与显示(Anthropic / Z.ai / MiniMax / Kimi / OpenRouter / SiliconFlow / CommandCode / SCNet / 火山方舟 九家,含 Volcano Ark AK/SK 签名) · Plan/API 双轨计费(订阅额度与按量金额分离统计,每 1% 额度与满窗的 token/等值金额估算及日/周/月曲线) · 输入框上方额度横条(预算/Go/Coding Plan 用量一条横排显示,可开关)
 
-[![version](https://img.shields.io/badge/version-1.5.42-4176E6)](https://github.com/Han-1413141/dsh-cost-meter)
+[![version](https://img.shields.io/badge/version-1.6.2-4176E6)](https://github.com/Han-1413141/dsh-cost-meter)
 [![npm](https://img.shields.io/npm/v/dsh-cost-meter?label=npm)](https://www.npmjs.com/package/dsh-cost-meter)
 [![license](https://img.shields.io/badge/license-MIT-green)](LICENSE)
 [![dsh](https://img.shields.io/badge/DeepSeek%20Harness-dsh--plugin-4176E6)](https://github.com/deepseek-ai/deepseek-harness)
@@ -29,13 +29,14 @@
 | 官方余额 | 侧边栏顶部 / 设置页(可配) | 总余额 / 赠送 / 充值,自动刷新 + 手动刷新;可选三段进度条(蓝/橙/灰),当日段只统计官方渠道费用(不含 Coding Plan / 自定义 Provider) |
 | 自定义 Provider 余额 | 侧边栏 / 设置页(可配) | 可配置 HTTP 查询任意 Provider 余额(LiteLLM 等);中/英名称、币种、extract 规则(点路径 / 数字常量 / add / subtract / divide,divide 适配 NewApi 等 quota 端点,见下方[示例](#自定义-provider-余额配置示例newapi-模板));与 Coding Plan 同区可折叠配置 |
 | OpenCode Go 额度 | 侧边栏 / 设置页 / 右下角(dock,可配) | 滚动 5 小时 / 本周 / 本月用量百分比与重置时间,三档可分别开关,可同时显示预算已用%;Key 自动发现(DSH 凭据库 OPENCODE_GO_API_KEY / 环境变量 / opencode 登录态)或手动填写 |
-| Coding Plan 额度 | 侧边栏 / 设置页(每家可配) | 多厂商 coding plan 订阅额度查询(Anthropic Claude Pro/Max、Z.ai/智谱 GLM、MiniMax Token Plan、Kimi Code 本周/5 小时配额(无订阅 Key 时回落 PAYG 余额)、OpenRouter credits、SiliconFlow 余额、CommandCode 5h/周窗口与月度 Credits 余额),各家独立启用开关、Key、显示位置与刷新间隔(侧边栏卡片与 Go 额度同款,收起窄栏显示百分比),凭据只发往官方端点;无凭据/无订阅为中性提示;SCNet 超算互联网 Token Plan 无 API 额度端点,按官方 Credits 抵扣表由本地账本估算月度用量(无需凭据) |
+| Coding Plan 额度 | 侧边栏 / 设置页(每家可配) | 多厂商 coding plan 订阅额度查询(Anthropic Claude Pro/Max、Z.ai/智谱 GLM、MiniMax Token Plan、Kimi Code 本周/5 小时配额(无订阅 Key 时回落 PAYG 余额)、OpenRouter credits、SiliconFlow 余额、CommandCode 5h/周窗口与月度 Credits 余额、火山方舟 Volcano Ark 5h/周/月三档(需 AK/SK 管控面 HMAC 签名,需 ArkReadOnlyAccess + BillingCenterReadOnlyAccess)),各家独立启用开关、凭据、显示位置与刷新间隔(侧边栏卡片与 Go 额度同款,收起窄栏显示百分比),凭据只发往官方端点;无凭据/无订阅为中性提示;SCNet 超算互联网 Token Plan 无 API 额度端点,按官方 Credits 抵扣表由本地账本估算月度用量(无需凭据) |
 | 额度横条 | 输入框上方(显示设置可开关) | 一条横排 chips 实时显示预算已用% / Go 主窗口 / 各已启用 Coding Plan 用量窗口(短标签+迷你进度条,≥80% 预警、≥100% 超支,悬停见重置时刻);点击任意 chip 即刷新对应数据源(budget→状态、Go→Go 额度、厂商→该家全部窗口),同一厂商多窗口融合为一条 chip 分段显示;首次更新弹引导卡由用户自主决定开关;无可用数据自动隐藏 |
 | 点击立即刷新 | 侧边栏余额/额度图框 | 官方余额 / 自定义余额 / Coding Plan 图框(含窄栏收起态)点击即触发一次查询,刷新中呼吸闪烁,失败保持原值并在悬停提示说明;键盘 Enter/Space 可触发;更新后首次进入有引导提示 |
 | 当日费用 | 侧边栏底部(设置按钮上方) | 「今日 ¥x」,悬停见调用次数与 token 明细 |
 | 预算图框 | 侧边栏底部(余额行与设置按钮之间) | 圆角方形图框:预算、已用%、进度条、今日费用与占预算%、已用/额度,≥80% 预警、≥100% 超支 |
 | 汇总卡片 | 设置页 | 今日 / 本月 / 累计费用与调用次数 |
 | Token 用量统计 | 设置页(费用设置) | 历史累计 token 总量(输入/缓存/输出/调用)+ 类 Codex 的 26 周每日用量方格热图,横向铺满设置页宽度,悬停见当日明细 |
+| Token Plan 用量统计 | 设置页(用量) | 各已启用 Coding Plan(含 Go)当前窗口的「每 1% 额度」与「满窗 100%」对应的 token 数与等值金额估算(采样差分/当前用量折算),附每日/每周/每月用量曲线;Plan 类调用金额只记等值,不动真金白银(issue #64) |
 | 今日会话明细 | 设置页 | 每个会话的调用次数、输入/缓存/输出 token 与费用 |
 | 历史记录 | 设置页 | 按天汇总,保留天数可配(默认 180 天) |
 | 历史按模型统计回填 | 设置页(按模型统计) | 按模型统计上线前的旧账本自动回放宿主会话日志重建逐模型 token/费用拆分(旧调用按当时基础价),日志已清理的部分归入「早期未分模型」残差行 |
@@ -50,6 +51,7 @@
 | AI 价格同步 | [提示词](docs/AI-PRICE-SYNC-PROMPT.md) | DeepSeek 官方同步;其他 provider 使用已核对的官方价格目录与手动配置 |
 | 模型与 Plan 适配说明 | [适配文档](docs/model-and-plan-adaptation.md) | 各厂商模型计费与 8 家 Coding Plan 的适配矩阵、自动匹配机制与价格来源([English](docs/model-and-plan-adaptation.en.md)) |
 | 峰/谷切换提醒图解 | [提醒文档](docs/peak-alert.md) | 峰谷切换前弹窗与系统通知的完整图解:效果截图(中/英)、设置项说明与使用建议([English](docs/peak-alert.en.md)) |
+| Token Plan 用量统计图解 | [面板文档](docs/token-plan-stats.md) | 每 1% 与满窗估算的四列含义、首尾差分估算方法与精度标注、口径边界(只统计 dsh 内调用)与用量曲线说明([English](docs/token-plan-stats.md#english)) |
 | 多 provider 计费 | 设置页 / 账本 | 支持 OpenAI、Anthropic、Google Gemini、Mistral 等 provider 的 input/output、缓存与 reasoning token 价格,按 provider+model 隔离计费 |
 | 模型名自动匹配 | 设置页 / 账本 | 未知模型 id 自动匹配价格表:忽略大小写/空格/横杠/点号与括号附注,归一化等价或请求名包含表内模型名即命中(如 `gpt5.6 luna(go)`);路由 provider(opencode/zen 等)下跨厂商全库查找;可关闭为仅精确;未命中模型可手动指定计费条目 |
 | 拓展价格表 | 设置页 → 拓展价格表 | 内置各厂商、按模型家族分类的参考价格目录(点开展开,厂商默认折叠);一键挂载参与计费,挂载的第三方模型默认收入表内可编辑;逐模型「在费用设置直接显示」开关自选哪些模型(含 DeepSeek)在「价格表」区直接显示 |
@@ -231,22 +233,22 @@
 dsh plugin --profile web add dsh-cost-meter
 ```
 
-**PowerShell 一键脚本**(复制整行粘贴回车;自动补齐 pnpm、自动探测 git,无需克隆仓库;安装链**固定到发布 tag `v1.5.42`**,建议先下载审阅再运行):
+**PowerShell 一键脚本**(复制整行粘贴回车;自动补齐 pnpm、自动探测 git,无需克隆仓库;安装链**固定到发布 tag `v1.6.2`**,建议先下载审阅再运行):
 
 ```powershell
-irm https://raw.githubusercontent.com/Han-1413141/dsh-cost-meter/v1.5.42/install.ps1 | iex
+irm https://raw.githubusercontent.com/Han-1413141/dsh-cost-meter/v1.6.2/install.ps1 | iex
 ```
 
 **或直接命令行**(机器上需已有 pnpm 与 git;同样固定到 tag):
 
 ```sh
-dsh plugin --profile web add github:Han-1413141/dsh-cost-meter#v1.5.42
+dsh plugin --profile web add github:Han-1413141/dsh-cost-meter#v1.6.2
 ```
 
 没有 git 时可用 GitHub tag 打包直链:
 
 ```sh
-dsh plugin --profile web add https://github.com/Han-1413141/dsh-cost-meter/archive/refs/tags/v1.5.42.tar.gz
+dsh plugin --profile web add https://github.com/Han-1413141/dsh-cost-meter/archive/refs/tags/v1.6.2.tar.gz
 ```
 
 安装后**重启** `dsh web`(插件行、Typert 清单与客户端 bundle 均在启动时扫描):
@@ -283,6 +285,9 @@ dsh plugin --profile web add link:./dsh-cost-meter  # 符号链接,改 lib/clien
 - 会话徽章与当日/月度/累计、预算一样,按每次调用的**实际时刻精确计费**(宿主导出的逐次成本);
 - 计费来源为每次模型调用的 usage 块(含子代理、压缩、标题等辅助调用),与账单口径一致;
 - 预算与超支提示**仅提醒,不阻止调用**。
+- **Plan/API 双轨计费**(issue #64):订阅制渠道(MiniMax / Codex 手动标记等)的调用金额只记「等值」,预算/今日费用/概览卡片等金额展示仅统计按量计费(API)部分;
+- **「含 Plan 总额」开关**:概览页汇总卡片下的快捷开关切换全部金额展示口径——关闭时仅计真金白银(API 渠道),开启后显示含 Plan 等值的总金额;网关路由(provider 缺失)调用的第三方目录模型自动归入对应订阅归类,无模型明细的历史残差计入 API 口径;设置 → 用量 的 Token Plan 统计面板提供每 1% 额度与满窗的 token/等值金额估算与日/周/月曲线;分类可在配置中按厂商或 provider:model 级覆盖。
+- **全仓安全审计修复(v1.6.2)**:账本退出丢写(close/flush 次序)、发版脚本命令注入、面板空指针崩溃(Go 月窗空值)3 项高危,及路由调用小时桶漏记、官方余额对账告警失效、自定义余额提取失败误显 $0、计费流中断泄漏等 30 余项中低危问题全量修复;账本损坏自动备份、写失败重试、配置补丁原子性、原型链与凭据处理加固。逐项清单见 [docs/security-audit-v1.6.2.md](docs/security-audit-v1.6.2.md)。
 
 ## 数据存储
 
@@ -355,6 +360,8 @@ dsh --profile web --port 3099                           # 真机启动(观察启
 - 价格同步会覆盖官方页面列出的同名模型价格,自定义模型条目不受影响;
 - 余额查询需要可访问 api.deepseek.com 的网络与有效 API Key;**API Key 只会发往官方域名**(baseURL 指向非官方域名时余额查询拒绝请求,模型请求不受影响);
 - OpenCode Go 额度接口为 opencode.ai 官方端点(社区文档);接口结构变化时设置页会显示错误,可在显示设置中关闭该显示;
+- Token Plan 统计的「每 1% 额度/满窗」为估算值:各家额度接口只返回百分比且读数存在个位级量化(显示 1% 的真实值可能在 0.5%~1.5%),插件以连续可信段的首尾差分推算以压低量化误差(跨度不足 5 个百分点时标注「读数精度受限」,样本超 7 天回退当前用量折算,窗口边界按小时对齐);服务端百分比统计的是该账号全部用量——同一 Key 在其它机器/CLI 的消耗不在本地账本,估算偏低属预期,仅供跨套餐横向比较;
+- Plan/API 双轨分类基于渠道与配置推断,混合订阅/按量使用同一厂商 Key 的场景(如 Kimi 订阅 + PAYG 混用)可在设置中按 provider:model 手动覆盖。
 - 安装/更新插件后需重启 `dsh web` 生效。
 
 ## 更新历史

package/docs/provider-pricing.json

@@ -1,17 +1,17 @@
 {
-  "generatedAt": "2026-08-18",
+  "generatedAt": "2026-08-25",
   "note": "Auto-generated from lib/pricing.js DEFAULT_PROVIDER_PRICE_TABLE. USD per 1M tokens; unpriced entries have no verified official price.",
   "providers": {
     "openai": {
       "models": {
         "gpt-5.6-sol": {
-          "input": 5,
-          "cachedInput": 0.5,
-          "output": 30,
+          "input": 2,
+          "cachedInput": 0.2,
+          "output": 10,
           "billingMode": "flat",
           "sourceUrl": "https://opencode.ai/docs/zen",
-          "checkedAt": "2026-08-17",
-          "notes": "≤272K 档;超过 272K 输入按 $10/$45 计;缓存写入 $6.25"
+          "checkedAt": "2026-08-25",
+          "notes": "≤272K 档;超过 272K 按 $4/$15 计(缓存读 $0.40、写入 $5);缓存写入 $2.50;目录标注 2026-09-18 前为五折促销价(issue #58)"
         },
         "gpt-5.6-terra": {
           "input": 2,
@@ -287,7 +287,7 @@
           "billingMode": "flat",
           "sourceUrl": "https://ai.google.dev/gemini-api/docs/pricing",
           "checkedAt": "2026-08-17",
-          "notes": "2026 年底前促销价;2027 起 $1.5/$7.5;Zen 网关按 $1.5/$7.5"
+          "notes": "2026 年底前促销价;2027 起 $1.5/$7.5;Go 网关按 $1.5/$7.5"
         },
         "gemini-3.6-flash": {
           "input": 1.5,
@@ -365,7 +365,7 @@
           "billingMode": "flat",
           "sourceUrl": "https://opencode.ai/docs/zen",
           "checkedAt": "2026-08-17",
-          "notes": "官方人民币价 ¥20/¥100/缓存¥2 每百万;此处为 Zen 网关美元价"
+          "notes": "官方人民币价 ¥20/¥100/缓存¥2 每百万;此处为 Go 网关美元价"
         },
         "kimi-k2.7-code": {
           "input": 0.95,
@@ -605,6 +605,30 @@
         }
       }
     },
+    "meta": {
+      "models": {
+        "muse-spark-1.2": {
+          "input": 1.25,
+          "cachedInput": 0.15,
+          "output": 4.25,
+          "billingMode": "flat",
+          "sourceUrl": "https://opencode.ai/docs/zen",
+          "checkedAt": "2026-08-25"
+        }
+      }
+    },
+    "meituan": {
+      "models": {
+        "longcat-2.0": {
+          "input": 0.3,
+          "cachedInput": 0.006,
+          "output": 1.2,
+          "billingMode": "flat",
+          "sourceUrl": "https://opencode.ai/docs/go",
+          "checkedAt": "2026-08-25"
+        }
+      }
+    },
     "nvidia": {
       "models": {
         "nvidia/nemotron-3-ultra-550b-a55b": {
@@ -647,9 +671,12 @@
           "checkedAt": "2026-08-17"
         },
         "glm-5.3": {
-          "unpriced": true,
+          "input": 1.4,
+          "cachedInput": 0.26,
+          "output": 4.4,
           "billingMode": "flat",
-          "notes": "Go 目录在册;官方未公布单价"
+          "sourceUrl": "https://opencode.ai/docs/go",
+          "checkedAt": "2026-08-25"
         },
         "glm-5.2": {
           "input": 1.4,
@@ -770,8 +797,25 @@
           "billingMode": "flat",
           "sourceUrl": "https://opencode.ai/docs/go",
           "checkedAt": "2026-08-17"
+        },
+        "longcat-2.0": {
+          "input": 0.3,
+          "cachedInput": 0.006,
+          "output": 1.2,
+          "billingMode": "flat",
+          "sourceUrl": "https://opencode.ai/docs/go",
+          "checkedAt": "2026-08-25"
+        },
+        "muse-spark-1.2-contributor": {
+          "input": 0.1,
+          "cachedInput": 0.002,
+          "output": 0.2,
+          "billingMode": "flat",
+          "sourceUrl": "https://opencode.ai/docs/go",
+          "checkedAt": "2026-08-25",
+          "notes": "低价换取提示词/补全用于训练 Meta 模型;限地区"
         }
       }
     }
   }
-}
+}

package/lib/backfill.js

@@ -14,7 +14,8 @@ import { readdirSync, readFileSync, statSync } from 'node:fs'
 import { join } from 'node:path'
 import * as zlib from 'node:zlib'
 import { costOf, providerPriceEntryFor, usdFromCost } from './pricing.js'
-import { localDayKey, zeroDay } from './store.js'
+import { localDayKey, zeroDay, splitLedgerApiCost } from './store.js'
+import { billingClassOf, enabledPlanSetOf } from './plan-billing.js'
 
 const ZSTD_MAGIC = 4247762216
 /** 打包行(文本/推理/工具调用增量游程)不含 header 与 usage,回放时跳过。 */
@@ -95,7 +96,10 @@ export function readSessionRecords(path) {
       const line = lines[i]
       if (line.length === 0) continue
       try {
-        records.push(JSON.parse(line))
+        const rec = JSON.parse(line)
+        // 打包行(text/reasoning/tool-call 增量游程)在解析入口就丢弃:它们是
+        // 大日志的体积主体,回放阶段反正会跳过,提前过滤才能兑现 OOM 缓解。
+        if (rec !== null && typeof rec === 'object' && !PACKED_ROW_TYPES.has(rec.type)) records.push(rec)
       } catch {
         // 坏行跳过:回放是尽力而为,不让单行损坏阻断整个会话。
       }
@@ -112,7 +116,9 @@ export function readSessionRecords(path) {
   }
   if (pending.length > 0) {
     try {
-      records.push(JSON.parse(pending))
+      const rec = JSON.parse(pending)
+      // 与上方 consumeText 同规则:打包行不入记录(文件末尾无换行的残行兜底)。
+      if (rec !== null && typeof rec === 'object' && !PACKED_ROW_TYPES.has(rec.type)) records.push(rec)
     } catch {
       // 文件末尾无换行的残行:同上,尽力而为。
     }
@@ -185,7 +191,7 @@ export function listSessionLogs(root, onlySessionIds = null) {
  * @returns { sessionId, title, createdAt, days, seedDays }。
  */
 export function replaySessionRecords(records, config, wantDates = null) {
-  const zeroBuckets = () => ({ input: 0, output: 0, cacheRead: 0, cacheWrite: 0, reasoning: 0, calls: 0, cost: 0 })
+  const zeroBuckets = () => ({ input: 0, output: 0, cacheRead: 0, cacheWrite: 0, reasoning: 0, calls: 0, cost: 0, apiCost: 0 })
   const num = value => {
     const n = Number(value)
     return Number.isFinite(n) && n > 0 ? n : 0
@@ -209,6 +215,7 @@ export function replaySessionRecords(records, config, wantDates = null) {
       reasoning: current.reasoning + sign * sample.buckets.reasoning,
       calls: current.calls + sign,
       cost: current.cost + sign * sample.cost,
+      apiCost: (current.apiCost ?? 0) + sign * (sample.apiCost ?? sample.cost),
     }
   }
   for (const event of records) {
@@ -242,14 +249,16 @@ export function replaySessionRecords(records, config, wantDates = null) {
     let usage = null
     let turn = 0
     let step = 0
-    if (event.type === 'assistant/chunk' && event.data?.chunk?.type === 'usage' && event.data.chunk.usage !== undefined) {
+    // 判空与去重键缺省值与 index.js 投影完全同构(usage null 不入账;turn/step
+    // 缺省 0,保证实时投影与回放对同一事件产生同一 (turn,step) 去重键)。
+    if (event.type === 'assistant/chunk' && event.data?.chunk?.type === 'usage' && event.data.chunk.usage != null) {
       usage = event.data.chunk.usage
-      turn = event.data.turn
-      step = event.data.step
-    } else if (event.type === 'assistant/message' && event.data?.usage !== undefined) {
+      turn = event.data.turn ?? 0
+      step = event.data.step ?? 0
+    } else if (event.type === 'assistant/message' && event.data?.usage != null) {
       usage = event.data.usage
-      turn = event.data.turn
-      step = event.data.step
+      turn = event.data.turn ?? 0
+      step = event.data.step ?? 0
     } else {
       continue
     }
@@ -289,11 +298,14 @@ export function replaySessionRecords(records, config, wantDates = null) {
     const cost = usdFromCost(priced,
       resolved.billingMode === 'deepseek-peak' && config?.prices?.currency === 'CNY' ? 'CNY' : 'USD',
       config?.exchangeRate)
+    // Plan/API 双轨分类(issue #64):与 store.account() 同一分类器,回放条目
+    // 的 apiCost 只含真金白银部分(plan 类金额仅记等值);传 prices 供路由判定。
+    const apiCost = billingClassOf(provider, model, config?.planBilling, enabledPlanSetOf(config), config?.prices) === 'api' ? cost : 0
     const providerKey = `${provider}:${model}`
     // 聚合目标按事件分段路由;被替换的旧样本从它当时所在的段扣回(与旧版
     // 单流先减后加的净效果一致,仅种子段的量落到 seedDays 而非 days)。
     if (prev !== null) shift(prev, -1, prev.seed ? seedDays : days)
-    const sample = { key, date, providerKey, buckets, cost, seed: isSeed }
+    const sample = { key, date, providerKey, buckets, cost, apiCost, seed: isSeed }
     shift(sample, 1, isSeed ? seedDays : days)
     last = sample
   }
@@ -398,9 +410,12 @@ export async function backfillLegacyLedger(ledger, sessionsRoot) {
       // 回放完整覆盖当日全部调用与 token 时,按回放结果重算当日总额(issue #18):
       // 旧版本曾把订阅制模型模糊匹配到同家族付费价实时误计费,回放按事件时刻
       // 正确计价,重算可修正历史虚高;仅部分覆盖时保留原始记录,差额入 legacy 行。
+      // reasoning 必须一并相等:旧样本缺失 reasoningTokens 时其余四桶仍可能恰好
+      // 相等,漏判会以不含 reasoning 成本的回放值覆盖账本且无痕。
       if (replayed.calls === (day.calls ?? 0)
         && replayed.input === (day.input ?? 0) && replayed.output === (day.output ?? 0)
-        && replayed.cacheRead === (day.cacheRead ?? 0) && replayed.cacheWrite === (day.cacheWrite ?? 0)) {
+        && replayed.cacheRead === (day.cacheRead ?? 0) && replayed.cacheWrite === (day.cacheWrite ?? 0)
+        && replayed.reasoning === (day.reasoning ?? 0)) {
         day.cost = replayed.cost
         result.recosted = (result.recosted ?? 0) + 1
       }
@@ -436,14 +451,16 @@ export async function backfillLegacyLedger(ledger, sessionsRoot) {
         acc.output += b.output ?? 0
         acc.cacheRead += b.cacheRead ?? 0
         acc.cacheWrite += b.cacheWrite ?? 0
+        acc.reasoning += b.reasoning ?? 0
         acc.calls += b.calls ?? 0
         acc.cost += b.cost ?? 0
         return acc
-      }, { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, calls: 0, cost: 0 })
+      }, { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, reasoning: 0, calls: 0, cost: 0 })
       if (sTotals.calls === (session.calls ?? 0)
         && dayIsEmpty
         && sTotals.input === (session.input ?? 0) && sTotals.output === (session.output ?? 0)
-        && sTotals.cacheRead === (session.cacheRead ?? 0) && sTotals.cacheWrite === (session.cacheWrite ?? 0)) {
+        && sTotals.cacheRead === (session.cacheRead ?? 0) && sTotals.cacheWrite === (session.cacheWrite ?? 0)
+        && sTotals.reasoning === (session.reasoning ?? 0)) {
         session.cost = sTotals.cost
       }
     }
@@ -467,7 +484,12 @@ export async function backfillLegacyLedger(ledger, sessionsRoot) {
       }
     }
   }
-  if (result.days > 0 || result.sessions > 0 || result.titles > 0) ledger.scheduleWrite()
+  // 双轨口径一致性(B-4):上面的金额重算只改写了容器 cost,apiCost 仍是旧值;
+  // 向下修正后会出现 apiCost > cost 的倒挂(净化过的旧条目默认 apiCost = cost,
+  // 一次性迁移标记不会自愈)。按最新桶级分类重建全部容器的 apiCost——
+  // store.splitLedgerApiCost 幂等且只读分类配置,store.js 不反向依赖本模块,无环。
+  const apiTouched = splitLedgerApiCost(ledger)
+  if (result.days > 0 || result.sessions > 0 || result.titles > 0 || apiTouched > 0) ledger.scheduleWrite()
   return result
 }
 
@@ -491,7 +513,7 @@ function mergeDayMaps(target, source) {
 
 function mergeBucketsInto(target, source) {
   for (const [key, b] of Object.entries(source)) {
-    const current = target[key] ?? { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, reasoning: 0, calls: 0, cost: 0 }
+    const current = target[key] ?? { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, reasoning: 0, calls: 0, cost: 0, apiCost: 0 }
     target[key] = {
       input: current.input + (b.input ?? 0),
       output: current.output + (b.output ?? 0),
@@ -500,6 +522,7 @@ function mergeBucketsInto(target, source) {
       reasoning: current.reasoning + (b.reasoning ?? 0),
       calls: current.calls + (b.calls ?? 0),
       cost: current.cost + (b.cost ?? 0),
+      apiCost: (current.apiCost ?? 0) + (b.apiCost ?? b.cost ?? 0),
     }
   }
 }
@@ -515,12 +538,13 @@ function cloneDayMap(pm) {
       reasoning: b.reasoning ?? 0,
       calls: b.calls ?? 0,
       cost: b.cost ?? 0,
+      apiCost: b.apiCost ?? b.cost ?? 0,
     }
   }
   return out
 }
 
-/** 汇总一组 provider:model 桶的合计(input/output/…/calls/cost)。 */
+/** 汇总一组 provider:model 桶的合计(input/output/…/calls/cost/apiCost)。 */
 function sumDayMap(pm) {
   return Object.values(pm).reduce((acc, b) => {
     acc.input += b.input ?? 0
@@ -530,8 +554,9 @@ function sumDayMap(pm) {
     acc.reasoning += b.reasoning ?? 0
     acc.calls += b.calls ?? 0
     acc.cost += b.cost ?? 0
+    acc.apiCost += b.apiCost ?? b.cost ?? 0
     return acc
-  }, { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, reasoning: 0, calls: 0, cost: 0 })
+  }, { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, reasoning: 0, calls: 0, cost: 0, apiCost: 0 })
 }
 
 /** 把合计桶累加到日期/会话条目的顶层字段上。 */
@@ -543,6 +568,7 @@ function addTotalsTo(target, totals) {
   target.reasoning += totals.reasoning
   target.calls += totals.calls
   target.cost += totals.cost
+  target.apiCost = (target.apiCost ?? 0) + (totals.apiCost ?? 0)
 }
 
 /** 由回放结果构造账本会话条目(title/at 缺席时不写键,与 Typert schema 一致)。 */
@@ -556,6 +582,7 @@ function sessionEntryOf(sessionId, info, totals, pm) {
     reasoning: totals.reasoning,
     calls: totals.calls,
     cost: totals.cost,
+    apiCost: totals.apiCost ?? totals.cost,
     byProviderModel: cloneDayMap(pm),
   }
   if (typeof info.title === 'string' && info.title.length > 0) entry.title = info.title
@@ -662,6 +689,9 @@ function subtractBucketInto(target, source) {
   target.reasoning = Math.max(0, (target.reasoning ?? 0) - (source.reasoning ?? 0))
   target.calls = Math.max(0, (target.calls ?? 0) - (source.calls ?? 0))
   target.cost = Math.max(0, (target.cost ?? 0) - (source.cost ?? 0))
+  if (target.apiCost !== undefined || source.apiCost !== undefined) {
+    target.apiCost = Math.max(0, (target.apiCost ?? target.cost ?? 0) - (source.apiCost ?? source.cost ?? 0))
+  }
 }
 
 /**
@@ -676,10 +706,11 @@ function subtractBucketInto(target, source) {
  * @param sessionsRoot - 宿主会话根目录($DSH_HOME/sessions)。
  * @returns { sessions, days, scanned } 受影响的 fork 会话数、扣除的日期数与扫描文件数。
  */
-export function repairForkSeed(ledger, sessionsRoot) {
+export async function repairForkSeed(ledger, sessionsRoot) {
   const result = { sessions: 0, days: 0, scanned: 0 }
   for (const path of listSessionLogs(sessionsRoot)) {
-    result.scanned += 1
+    // 全量同步扫描会长时间占住事件循环:每 8 份让出一次(与回填/导入同策略)。
+    if ((result.scanned += 1) % 8 === 0) await new Promise(resolve => setImmediate(resolve))
     let records
     try {
       records = readSessionRecords(path)
@@ -748,7 +779,7 @@ export function repairForkSeed(ledger, sessionsRoot) {
  * @param ledger - 已打开的账本。
  * @returns { groups, removedCost } 合并的重复组数与扣除的金额合计(USD)。
  */
-export function repairProviderDupes(ledger) {
+export async function repairProviderDupes(ledger) {
   const result = { groups: 0, removedCost: 0 }
   // 清洗单个容器(day 或 session):返回是否发生扣除。
   const repairContainer = (container) => {
@@ -781,7 +812,10 @@ export function repairProviderDupes(ledger) {
     return touched
   }
   let scheduled = false
+  let scannedContainers = 0
   for (const day of Object.values(ledger.days ?? {})) {
+    // 大账本逐容器指纹分组是纯 CPU 工作:每 8 个容器让出一次事件循环。
+    if ((scannedContainers += 1) % 8 === 0) await new Promise(resolve => setImmediate(resolve))
     if (day === null || typeof day !== 'object') continue
     if (repairContainer(day)) scheduled = true
     if (Array.isArray(day.sessions)) {

package/lib/billing-stream.js

@@ -17,7 +17,15 @@
 
 import { AsyncLocalStorage } from 'node:async_hooks'
 
-/** 嵌套深度标记:值恒为 1(存在即嵌套);按异步上下文隔离,并发流互不串扰。 */
+/**
+ * 嵌套深度标记:值恒为 1(存在即嵌套);按异步上下文隔离,并发流互不串扰。
+ *
+ * 设计边界(R-7):恒定深度标记会把「外层 pull 的异步上下文内同步分发瀑布时」
+ * 发起的任何 ctx.llm.stream() 一律判定为已记账的嵌套调用。单层包装链
+ * (issue #48 目标:适配器在自己 stream() 体内恰好发起一次上游调用)语义正确;
+ * 若未来出现扇出型路由——单次 pull 内派生 >1 条上游流——第二条及之后的上游
+ * 将被误判为嵌套而漏记。出现此类路由时需改用按流实例区分的标记身份。
+ */
 const llmStreamDepth = new AsyncLocalStorage()
 
 /**
@@ -37,6 +45,7 @@ export function createLlmStreamBilling({ account }) {
     return (async function* costMeterStream() {
       let usage = null
       const iterator = downstream[Symbol.asyncIterator]()
+      let completed = false
       try {
         for (;;) {
           // 在深度标记内拉取下游:下游适配器(modlens 等)体内再发起的
@@ -44,12 +53,21 @@ export function createLlmStreamBilling({ account }) {
           const result = await llmStreamDepth.run(1, () => iterator.next())
           if (result.done) break
           const chunk = result.value
-          if (chunk !== null && chunk !== undefined && chunk.type === 'usage' && chunk.usage !== undefined) {
+          if (chunk !== null && chunk !== undefined && chunk.type === 'usage' && chunk.usage != null) {
             usage = chunk.usage
           }
           yield chunk
         }
+        completed = true
       } finally {
+        // 手写 for(;;) 不具备 for-await 的自动关闭语义:消费方提前中断(break/
+        // return/throw)时必须显式向下游传播 return(),否则上游 HTTP 流会继续
+        // 跑完(供应商照常计满 token,usage 块却无人读取),socket 悬挂泄漏。
+        if (!completed) {
+          try {
+            await iterator.return?.()
+          } catch {}
+        }
         if (usage !== null) {
           try {
             account(usage, options?.model, options?.sessionId, Date.now(), options?.provider)