dsh-connect-workbuddy 5.2.1 → 5.2.3

package/CHANGELOG.md

@@ -1,5 +1,41 @@
 # Changelog
 
+## 5.2.3 (2026-10-22)
+
+### Fixes
+
+- **「一键测速启用模型」按钮永远点不动（5.2.0 引入，截图发现）** —— 那个按钮的 `disabled` 条件里写的是 `tpsProbing !== undefined`，而 `tpsProbing` 是 `useState<Record<string, boolean>>({})`：**一个永远是对象的记录，永远不会 `undefined`**。于是比较恒为真，按钮永久禁用，整个测速功能从界面上根本进不去，而且没有任何提示说明为什么。
+  - **为什么会写错**：旁边两个可用性按钮比较的是 `sweeping !== undefined`，在那个变量上是对的 —— `sweeping` 真的是 `{done,total} | undefined`，`undefined` 确实表示「没在跑」。我照抄了那个表达式的形状，却没带上让它成立的事实：不同的状态变量，可取的值集合不同。**对一个 map 来说，「忙」是「有键」，不是「已定义」。**
+  - 修法：新增派生的 `tpsBusy = Object.keys(tpsProbing).length > 0`，用它替代恒真的比较。
+
+### Tests
+
+- `tests/card-tps-button.spec.tsx`（5 条，新增）：**断言的是 DOM 上真实的 `disabled` 布尔值，不是 JSX 文本** —— bug 就住在那条表达式里，文本级断言会和它一起说谎。
+  - 有启用模型且没在跑时**不可禁用**（这一条在修复前必红）
+  - 点击后真的发出请求，且只测启用的模型
+  - **一次只测一个模型**（路由的上限，防止一次点击变成对整份名册的无账花费）
+  - **一个都没启用时确实禁用**（防止修复把按钮变成无条件可点）
+  - 点击后**行上真的出现数字**（只发请求不渲染的按钮会骗过前四条）
+  - 已验证：把修复临时撤回，其中 4 条立刻转红。
+- 全量：**112 文件 / 1812 条全绿**；两个 tsconfig 均 exit 0；build 成功。
+
+## 5.2.2 (2026-10-22)
+
+### Fixes
+
+- **排队不再承诺、也不再触发自动重试（按维护者要求）** —— 两半都得改，因为它们是同一件事。
+  - **文案**：原来结尾是「约 N 秒后自动重试」。**没有任何东西在重试**：插件没有排队轮询循环（`src/queue.ts` 的 "WHAT THIS MODULE DOES NOT DO YET" 写明了这一点），也不会重发。现在改成「请稍后重新发送这条消息，或换一个模型」——位次照常报，但把决定权交回给用户。
+  - **请求本身**：那句谎言还有成本。`@deepseek-ai/dsh-llm` 把 HTTP 429 映射成 `RATE_LIMIT`，而 `RATE_LIMIT` 在它的 `DEFAULT_RETRYABLE_CODES` 里 —— 所以宿主会跑 **5 次** 原地重试（0.5s→8s 退避，实测合计约 15 秒）。一次排队在几千位之后，15 秒不可能排到，这 5 次全是**真实请求**，每次都只为再听一遍同一个位次。现在该响应带上 `x-should-retry: false`（两个客户端都**先**看这个头，再看状态码）与 `retry-after: 0`。
+  - **不转发上游的 `retry_after`**：7 秒也清不掉 9021 位的队，转发它只会让客户端先等一会再收到同一个答复。
+  - **`retryAfterSec` 字段一并删除**：它只用来喂那句倒计时，现在没人读。留着解析但不用的数据，等于给这个承诺留了条回头的路。
+  - **其他 429 仍然可以重试**：`model_busy`（14003）是模型级限流，确实约一分钟就恢复，宿主重试正是对的处置。只把排队这一类标成终局，不是一刀切。
+
+### Tests
+
+- `tests/queue-no-retry.spec.ts`（6 条，新增）：`x-should-retry: false`、`retry-after: 0`、不转发上游 `retry_after`、位次与队列长度仍在、文案里没有「自动重试」、**其他 429 不受影响**（最后这条防止把可恢复的限流也变成即时失败）。
+- `tests/queue.spec.ts`：`retryAfterSec` 相关断言改写为「不携带倒计时」。
+- 全量：**111 文件 / 1807 条全绿**；两个 tsconfig 均 exit 0；build 成功。
+
 ## 5.2.1 (2026-10-22)
 
 ### Fixes

package/README.md

@@ -43,7 +43,7 @@
 - **模型管理** —— 从上游刷新完整模型目录，逐项勾选启用或禁用；刷新是草稿操作，点保存才生效。上游同时给出积分倍率、上下文/输出上限与推理档位；**图片输入按厂商核验表判定**：刷新时只有厂商文档明确写为原生多模态的模型才自动勾选，文档写明纯文本的不勾，**没有厂商资料的一律按「未核实」不猜**（上游的 `supportsImages` 是平台侧声明，不等于模型能看图，不参与判定）。用户仍可随时手动勾选，勾选状态保存在本地。**每个模型行还有「可关闭思考」（`off`）勾选框**：决定要不要为该模型提供「完全关闭思考」这一档位。默认由内置复核名单给出（上游声称可关闭、实测却对它返回 HTTP 400 的 9 个模型默认不勾），你可以按模型改勾/取消——**取消**用于当前不需要该档位的模型；**名单内那 9 个模型不可勾选**（行上标「拒绝 off」，悬停有说明）：该档位被勾上后不只是多一个选项，pi-ai 会把它当作**所有未指定档位请求的默认值**，压缩摘要请求正属于这一类，勾一次等于让该模型的每一次 `/compact` 都 400 / 11133（issue #83）。名单外的模型仍完全由你的勾选与自身声明决定；首次保存即写入你的勾选，之后不再重算。
 - **模型行上的活动徽标** —— 上游的模型目录里带着活动信息（免费试用、夜间折扣、限时免费、订阅优先等），插件**不新增任何请求**（复用刷新模型时本来就会拉的那份目录配置），直接在模型行上画一枚胶囊徽标：颜色与文案都由上游给出，悬停能看到上游的活动说明、活动的起止日期与每日生效时段（例如「夜间 23:00–7:50」）。**徽标是「此刻」的函数**：活动只在它自己声明的时间窗内显示，窗口一关徽标就自己消失；跨零点的时段按上游声明的时区判断；同一个模型命中多条活动时只画优先级最高的那条。上游没给结束时间的权益标记（如「订阅优先」）则常驻，直到上游把它下架。**徽标只是提示，不改变模型行为**——它标注的是上游此刻给出的活动与折扣，不替你做选择。（两个区域的上游目录都下发活动字段：国内版按 App 身份读 `/v3/config` 时带 8 条促销 + 1 条等级标记，国际版带 5 条 + 1 条；取证与判定规则见 [docs/PROMO_BADGE.md](docs/PROMO_BADGE.md)。）
 - **模型行上的繁忙程度** —— 上游另有一条只读探针（`POST /v2/chat/busy`，一次最多 50 个 id，免费、不动账号状态）回答「这个模型此刻多忙」，插件在每次刷新目录时顺带问一次，把等级拼在倍率后面：`Hy4 preview · x0.29 · 极度拥堵`。**只画 `busy` / `congested` / `unavailable`**（轻微拥堵 / 极度拥堵 / 不可用）：`idle` 有词但不画——实测 20 个模型里 7 个空闲，每行都挂就成了装饰而不是警告；`normal` 与探针没回答的 `unknown` 一律不画，**「没被告知」不会被当成「一切正常」**。**等级不落盘**：目录是能活过重启的事实，负载只是那一分钟的事实，所以等级只盖在运行时目录上、随进程消失，探针失败只记一条 warn，绝不挡住目录刷新。实测这条探针在 CLI 与 App 两种身份下答案逐字节相同，50 个 id 的上限也真实存在（62 个 → HTTP 400），19 个模型 186 ms 返回。
-- **排队时说清楚你排在第几位** —— 上游用三个业务码表达排队（`6020` 等待、`6021` 队列满、`6022` 已达上限）。撞上它们时，插件不再给一句上游报错，而是「模型排队中：你排在第 8777 位，队列共 9021 个请求；约 7 秒后自动重试」。**排队不会触发换号**：队列跟随的是模型与平台级的队伍，换个账号拿到的还是同一个位次，所以这里既不走账号池也不烧额外请求（上游 `/v2/chat/queue/status` 回的 `message` 是 i18n key 而不是人话，插件因此从不转述它，只用自己的措辞）。**「排到之后自动重发」这一半没有实现**——它需要排队答案在 OpenAI 兼容通道上的真实信封，没测到就不写。
+- **排队时说清楚你排在第几位** —— 上游用三个业务码表达排队（`6020` 等待、`6021` 队列满、`6022` 已达上限）。撞上它们时，插件不再给一句上游报错，而是「模型排队中：你排在第 8777 位，队列共 9021 个请求。请稍后重新发送这条消息，或换一个模型」。**排队不会触发换号**：队列跟随的是模型与平台级的队伍，换个账号拿到的还是同一个位次，所以这里既不走账号池也不额外烧请求（上游 `/v2/chat/queue/status` 回的 `message` 是 i18n key 而不是人话，插件因此从不转述它，只用自己的措辞）。**排队不会自动重发，也不会被自动重试**——插件没有排队轮询循环（它需要排队答案在 OpenAI 兼容通道上的真实信封，没测到就不写），并且明确告诉宿主不要重试（`x-should-retry: false`）：一次排队在几千位之后，重试十几次也轮不到，只会把真实请求白白花掉。位次报出来，由你决定是重发还是换模型。
 - **模型可用性测试（含冷却恢复时间）** —— 限量模型的冷却时间过去只能靠反复试错去猜。现在每个模型行有一个「测试」按钮，直接告诉你它此刻能否使用、不能用时何时恢复。**它发的是一条真实体积的请求（约 25k 输入 token）**，而不是「发个 ping 看通不通」——因为上游的限流**按请求体积触发**（实测约 20k 能过、约 30k 被拒），小请求看着正常、长对话里却可能每次都被拒，那正是「测试通过、实际被限」的来源。**恢复时间的来源**：上游的 429 **不带** `Retry-After` 响应头（也没有 `X-RateLimit-*`），但它把重置时间写在**响应体**里——`{"code":6004,"msg":"…将在 2026-09-30 02:30:30 UTC+8 重置…"}`。插件从响应体解析该时间（并按其声明的 UTC 偏移换算，不受你本机时区影响），因此卡片能直接显示「✗ 被限流 · 2026/9/30 02:30:30 之后可再用」。三处来源都没说时，插件**不编造倒计时**，如实显示「上游未给出何时恢复」。限流**不会**去借月周期包的刷新时间（那会让人为几秒的限流去等几周）。探测在后台发起，**凭据不会传到页面**；结果不写入设置，也不改变你已保存的勾选。
 
   **两点需要知道**：① **它真的花钱**——实测一次报 `credit` 0.02（被限流时直接 429，几乎不花）。因此**只能按单个模型手工点**，没有「测试已勾选」这类批量按钮：一次覆盖整个列表就是一次不可控支出（服务端也据此**只接受单个模型**，多模型请求会被拒）。② 上游虽宣称 1M 上下文，但节流线远低于此；插件按模型的**原生窗口**向 DSH 宣称（3.5.0 起不再默认压到 200K），所以 DSH 很晚才压缩对话——长会话撞限流是这条机制的直接结果。想让 DSH 更早压缩，就在模型行上手动设一个「DSH 上下文预算」上限（可选 200K / 500K / 模型原生窗口）。

package/lib/bin.js

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-import { C as defaultDesktopAuthCandidates, Ct as findWorkbuddyAppExecutableWithSource, L as WorkBuddyUpstreamClient, M as workbuddyOwnAuthPath, _t as WORKBUDDY_APP_EXECUTABLE_ENV, a as readHostHeartbeat, b as WorkBuddyCredentialStore, c as WORKBUDDY_CONNECT_VERSION, d as FALLBACK_WORKBUDDY_MODELS_GLOBAL, jt as hostPeersInfo, k as legacyWorkbuddyOwnAuthPath, l as piAiRuntimeInfo, o as workbuddyHostHeartbeatPath, r as isHeartbeatProcessAlive, u as FALLBACK_WORKBUDDY_MODELS, w as defaultDesktopAuthDirs, y as WORKBUDDY_AUTH_FILE_ENV } from "./host-heartbeat-Icz67WtG.js";
+import { C as defaultDesktopAuthCandidates, Ct as findWorkbuddyAppExecutableWithSource, L as WorkBuddyUpstreamClient, M as workbuddyOwnAuthPath, _t as WORKBUDDY_APP_EXECUTABLE_ENV, a as readHostHeartbeat, b as WorkBuddyCredentialStore, c as WORKBUDDY_CONNECT_VERSION, d as FALLBACK_WORKBUDDY_MODELS_GLOBAL, jt as hostPeersInfo, k as legacyWorkbuddyOwnAuthPath, l as piAiRuntimeInfo, o as workbuddyHostHeartbeatPath, r as isHeartbeatProcessAlive, u as FALLBACK_WORKBUDDY_MODELS, w as defaultDesktopAuthDirs, y as WORKBUDDY_AUTH_FILE_ENV } from "./host-heartbeat-CgYyUwpM.js";
 import { realpathSync } from "node:fs";
 import { fileURLToPath } from "node:url";
 //#region src/bin.ts

package/lib/client.js

@@ -6139,6 +6139,17 @@ window.__ModuleLoader__.load({
 				}
 			};
 			/**
+			* Whether ANY speed run is in flight, for the sweep button's disabled state.
+			*
+			* NOT `tpsProbing !== undefined`. `tpsProbing` is a record that is ALWAYS an
+			* object, so that comparison is permanently true and the button was
+			* permanently disabled — the defect this corrects. The other two sweep buttons
+			* compare `sweeping !== undefined` because `sweeping` really is an
+			* `{done,total} | undefined` pair; this map has no such absent state, so the
+			* emptiness of its keys is what "busy" means.
+			*/
+			const tpsBusy = Object.keys(tpsProbing).length > 0;
+			/**
 			* Sweep speed for the models this region has ENABLED.
 			*
 			* Enabled-only rather than the whole catalog: a speed run is three real
@@ -6614,7 +6625,7 @@ window.__ModuleLoader__.load({
 														/* @__PURE__ */ (0, react_jsx_runtime.jsx)("button", {
 															type: "button",
 															className: "dsm-btn dsm-btn-outline dsm-workbuddy-models-sweep",
-															disabled: busy || tpsSweeping !== void 0 || tpsProbing !== void 0 || activeEnabledIds.size === 0,
+															disabled: busy || tpsSweeping !== void 0 || tpsBusy || activeEnabledIds.size === 0,
 															title: t("row.tpsSweepEnabledHint"),
 															onClick: sweepTpsEnabledModels,
 															children: tpsSweeping === void 0 ? t("row.tpsSweepEnabled") : t("row.tpsTesting", {

package/lib/{host-heartbeat-Icz67WtG.js → host-heartbeat-CgYyUwpM.js}

@@ -1045,8 +1045,6 @@ const WORKBUDDY_QUEUE_CODES = [
 	6021,
 	6022
 ];
-/** The app's default when the upstream sends no `retry_after` (`?? 5`). */
-const WORKBUDDY_QUEUE_DEFAULT_RETRY_SEC = 5;
 /** True when a business code is one of the app's three queue codes. */
 function isQueueCode(value) {
 	return typeof value === "number" && WORKBUDDY_QUEUE_CODES.includes(value);
@@ -1058,12 +1056,10 @@ function reportFrom(code, data) {
 	const numberField = (key) => typeof body[key] === "number" && Number.isFinite(body[key]) ? body[key] : void 0;
 	const position = numberField("queue_position");
 	const size = numberField("queue_size");
-	const retryAfter = numberField("retry_after");
 	return {
 		code,
 		...position === void 0 ? {} : { position },
-		...size === void 0 ? {} : { size },
-		retryAfterSec: retryAfter !== void 0 && retryAfter > 0 ? retryAfter : WORKBUDDY_QUEUE_DEFAULT_RETRY_SEC
+		...size === void 0 ? {} : { size }
 	};
 }
 /**
@@ -1121,9 +1117,22 @@ function readEnvelope$1(value) {
 *
 * `waiting` names both numbers because they answer different questions: the
 * POSITION is how long the wait is, the SIZE is whether it is moving at all.
+*
+* IT NO LONGER PROMISES "自动重试" (removed 2026-10-22, at the maintainer's
+* request). The sentence used to end with 「约 N 秒后自动重试」, and nothing
+* retried: this plugin does not poll `/v2/chat/queue/status` and does not resend
+* (see "WHAT THIS MODULE DOES NOT DO YET" above). The only retry that ever
+* happened was the HOST's own bounded in-place retry on a 429 — two attempts,
+* 0.5s then 1s, about 1.5 seconds — which is not "N seconds" and gives up long
+* before a real queue of thousands moves.
+*
+* So the sentence now states the WAIT and leaves the next move to the reader.
+* Reporting the position stays honest and useful — knowing you are 8,777th is
+* the answer to "wait or switch" — but claiming a recovery the plugin does not
+* perform turned a diagnostic into a promise it could not keep.
 */
 function queueNoticeText(report) {
-	if (report.code === 6020) return `模型排队中：你排在第 ${report.position === void 0 ? "未知" : `${report.position}`} 位${report.size === void 0 ? "" : `，队列共 ${report.size} 个请求`}；约 ${report.retryAfterSec} 秒后自动重试`;
+	if (report.code === 6020) return `模型排队中：你排在第 ${report.position === void 0 ? "未知" : `${report.position}`} 位${report.size === void 0 ? "" : `，队列共 ${report.size} 个请求`}。请稍后重新发送这条消息，或换一个模型`;
 	if (report.code === 6021) return `模型队列已满${report.size === void 0 ? "" : `（共 ${report.size} 个请求排队）`}，现在无法排队；请稍后重试或换一个模型`;
 	return `你的排队任务已达上限${report.size === void 0 ? "" : `（共 ${report.size} 个请求排队）`}，请等待已排队请求完成后再试`;
 }
@@ -5852,7 +5861,7 @@ function generationOf(version) {
 //#endregion
 //#region src/version.ts
 /** The npm package version this build was produced from. */
-const WORKBUDDY_CONNECT_VERSION = "5.2.1";
+const WORKBUDDY_CONNECT_VERSION = "5.2.3";
 //#endregion
 //#region src/host-heartbeat.ts
 /**

package/lib/index.js

@@ -1,4 +1,4 @@
-import { $ as regionOf, A as parseWorkBuddyAuth, At as workbuddyAppExecutableCandidates, B as forceNeutralHead, C as defaultDesktopAuthCandidates, D as isEncryptedCredentialError, Dt as macosNestedAppBundles, E as hasEncryptedCredentialFields, Et as macosBundleExecutable, F as WORKBUDDY_FALLBACK_SYSTEM_PROMPT, G as knownRegionOf, H as isConversationIdGateBody, I as WorkBuddyCredentialRejectedError, J as parseCreditMultiplier, K as modelIdOfBody, L as WorkBuddyUpstreamClient, M as workbuddyOwnAuthPath, Mt as resolveDshHome, N as CONTEXT_OVERFLOW_MARKER, Nt as withFileLock, O as isVaultAccountId, Ot as openEncryptedField, P as CREDENTIAL_REJECTED_CODE, Pt as writeFileAtomic, Q as prepareChatBody, R as classifyUpstreamError, S as authFileName, St as findWorkbuddyAppExecutable, T as defaultDesktopAuthPath, Tt as isWorkbuddyBundle, U as isCredentialRejectedError, V as isContextTooLongBody, W as isFirstMessageGateBody, X as parseUpstreamModel, Y as parseReasoning, Z as perturbChannelFingerprints, _ as ENCRYPTED_CREDENTIAL_CODE, _t as WORKBUDDY_APP_EXECUTABLE_ENV, a as readHostHeartbeat, at as offDefaultFor, b as WorkBuddyCredentialStore, bt as deriveAtRestKeyId, c as WORKBUDDY_CONNECT_VERSION, ct as withEffectiveOff, d as FALLBACK_WORKBUDDY_MODELS_GLOBAL, dt as pickImageModel, et as attachBusyLevels, f as WorkBuddyCatalog, ft as parseQueueAnswer, g as withoutRouters, gt as nativeModalityOf, h as fallbackModelsFor, ht as imageDefaultFor, i as processStartTimeMs, it as lastRefusalFor, j as workbuddyAccountId, k as legacyWorkbuddyOwnAuthPath, kt as readAtRestKey, l as piAiRuntimeInfo, lt as DEFAULT_IMAGE_SIZE, m as deriveCatalog, mt as NATIVE_MODALITY_BY_MODEL_ID, n as clearHostHeartbeat, nt as REFUSES_OFF_MODEL_IDS, o as workbuddyHostHeartbeatPath, ot as offRefused, p as applyContextBudgets, pt as queueNoticeText, q as openingEvidenceOfBody, r as isHeartbeatProcessAlive, rt as defaultOffModelIds, s as writeHostHeartbeat, st as recordRefusal, t as WORKBUDDY_HOST_HEARTBEAT_FILENAME, tt as busyDisplaySuffix, u as FALLBACK_WORKBUDDY_MODELS, ut as WorkBuddyImageError, v as WORKBUDDY_AUTH_FILENAME, vt as clearAtRestKeyCache, w as defaultDesktopAuthDirs, wt as isEncryptedFieldWrapper, x as WorkBuddyEncryptedCredentialError, xt as fetchAtRestKeyPayload, y as WORKBUDDY_AUTH_FILE_ENV, yt as deriveAtRestKey, z as declaredTools } from "./host-heartbeat-Icz67WtG.js";
+import { $ as regionOf, A as parseWorkBuddyAuth, At as workbuddyAppExecutableCandidates, B as forceNeutralHead, C as defaultDesktopAuthCandidates, D as isEncryptedCredentialError, Dt as macosNestedAppBundles, E as hasEncryptedCredentialFields, Et as macosBundleExecutable, F as WORKBUDDY_FALLBACK_SYSTEM_PROMPT, G as knownRegionOf, H as isConversationIdGateBody, I as WorkBuddyCredentialRejectedError, J as parseCreditMultiplier, K as modelIdOfBody, L as WorkBuddyUpstreamClient, M as workbuddyOwnAuthPath, Mt as resolveDshHome, N as CONTEXT_OVERFLOW_MARKER, Nt as withFileLock, O as isVaultAccountId, Ot as openEncryptedField, P as CREDENTIAL_REJECTED_CODE, Pt as writeFileAtomic, Q as prepareChatBody, R as classifyUpstreamError, S as authFileName, St as findWorkbuddyAppExecutable, T as defaultDesktopAuthPath, Tt as isWorkbuddyBundle, U as isCredentialRejectedError, V as isContextTooLongBody, W as isFirstMessageGateBody, X as parseUpstreamModel, Y as parseReasoning, Z as perturbChannelFingerprints, _ as ENCRYPTED_CREDENTIAL_CODE, _t as WORKBUDDY_APP_EXECUTABLE_ENV, a as readHostHeartbeat, at as offDefaultFor, b as WorkBuddyCredentialStore, bt as deriveAtRestKeyId, c as WORKBUDDY_CONNECT_VERSION, ct as withEffectiveOff, d as FALLBACK_WORKBUDDY_MODELS_GLOBAL, dt as pickImageModel, et as attachBusyLevels, f as WorkBuddyCatalog, ft as parseQueueAnswer, g as withoutRouters, gt as nativeModalityOf, h as fallbackModelsFor, ht as imageDefaultFor, i as processStartTimeMs, it as lastRefusalFor, j as workbuddyAccountId, k as legacyWorkbuddyOwnAuthPath, kt as readAtRestKey, l as piAiRuntimeInfo, lt as DEFAULT_IMAGE_SIZE, m as deriveCatalog, mt as NATIVE_MODALITY_BY_MODEL_ID, n as clearHostHeartbeat, nt as REFUSES_OFF_MODEL_IDS, o as workbuddyHostHeartbeatPath, ot as offRefused, p as applyContextBudgets, pt as queueNoticeText, q as openingEvidenceOfBody, r as isHeartbeatProcessAlive, rt as defaultOffModelIds, s as writeHostHeartbeat, st as recordRefusal, t as WORKBUDDY_HOST_HEARTBEAT_FILENAME, tt as busyDisplaySuffix, u as FALLBACK_WORKBUDDY_MODELS, ut as WorkBuddyImageError, v as WORKBUDDY_AUTH_FILENAME, vt as clearAtRestKeyCache, w as defaultDesktopAuthDirs, wt as isEncryptedFieldWrapper, x as WorkBuddyEncryptedCredentialError, xt as fetchAtRestKeyPayload, y as WORKBUDDY_AUTH_FILE_ENV, yt as deriveAtRestKey, z as declaredTools } from "./host-heartbeat-CgYyUwpM.js";
 import z from "@deepseek-ai/schemastery";
 import { appendFile, mkdir, readFile, writeFile } from "node:fs/promises";
 import { randomBytes, randomUUID, timingSafeEqual } from "node:crypto";
@@ -2434,11 +2434,12 @@ const KIND_HOST_CODE = { soft_rate: "quota_exceeded" };
 function hostErrorCode(kind) {
 	return KIND_HOST_CODE[kind] ?? kind;
 }
-function writeJson(res, status, body) {
+function writeJson(res, status, body, headers) {
 	const payload = JSON.stringify(body);
 	res.writeHead(status, {
 		"Content-Type": "application/json",
-		"Content-Length": Buffer.byteLength(payload)
+		"Content-Length": Buffer.byteLength(payload),
+		...headers
 	});
 	res.end(payload);
 }
@@ -2450,13 +2451,38 @@ function writeJson(res, status, body) {
 * must not be disguised as one. The upstream failure path passes an explicit
 * `code` so the host classifies it correctly — see {@link KIND_HOST_CODE}.
 */
-function writeOpenAIError(res, status, kind, message, code) {
+function writeOpenAIError(res, status, kind, message, code, headers) {
 	writeJson(res, status, { error: {
 		message,
 		type: kind,
 		code: code ?? kind
-	} });
+	} }, headers);
 }
+/**
+* Tell every retrying client NOT to retry this failure (2026-10-22, at the
+* maintainer's request: the queue's automatic retry is removed).
+*
+* `x-should-retry` is a non-standard header that both clients in this stack
+* check BEFORE their status-based rules — the pinned OpenAI SDK's `shouldRetry`
+* and pi-ai's `isRetryableProviderError` each return false on "false".
+*
+* WHY THE QUEUE MUST NOT BE RETRIED. Without this header a 429 is retried by
+* construction: `@deepseek-ai/dsh-llm` maps status 429 to the `RATE_LIMIT` code
+* and `RATE_LIMIT` is in `DEFAULT_RETRYABLE_CODES`, so the HOST runs five
+* in-place attempts with 0.5s→8s backoff. Measured, that budget totals roughly
+* 15 seconds — against a queue the upstream reports in thousands. Every one of
+* those attempts is a real request spent to be told the same position again, and
+* none of them can clear the queue.
+*
+* So the queue is now reported ONCE, with the position intact, and the reader
+* decides whether to resend or switch models. That is exactly what
+* {@link queueNoticeText} tells them to do — the two changes are one decision: a
+* diagnostic that declares itself final must actually be final.
+*/
+const NO_RETRY_HEADERS = Object.freeze({
+	"x-should-retry": "false",
+	"retry-after": "0"
+});
 /** Read a request body with a size cap; over-limit bodies fail the request. */
 function readBody(req) {
 	return new Promise((resolve, reject) => {
@@ -3570,7 +3596,7 @@ function createWorkBuddyShim(options) {
 			if (refusedReport !== void 0) reportRefusedTier(refusedReport);
 			const queue = result.kind === "model_queue" ? parseQueueAnswer(result.message) : void 0;
 			if (queue !== void 0) {
-				writeOpenAIError(res, 429, "model_queue", `${queueNoticeText(queue)}（上游 code ${queue.code}）${note}`);
+				writeOpenAIError(res, 429, "model_queue", `${queueNoticeText(queue)}（上游 code ${queue.code}）${note}`, void 0, NO_RETRY_HEADERS);
 				return;
 			}
 			if (result.kind === "policy_reject") {

package/package.json

@@ -2,7 +2,7 @@
   "name": "dsh-connect-workbuddy",
   "displayName": "DSH Connect WorkBuddy",
   "description": "把本机登录的 WorkBuddy 模型（国内版与国际版）接入 DeepSeek Harness：每次刷新按上游目录自动读取每个模型的积分倍率，支持逐模型图片输入开关、多账号切换与账号池自动换号，并提供每日签到、卡片与输入框里的胶囊积分读数等只读积分概览。",
-  "version": "5.2.1",
+  "version": "5.2.3",
   "author": "dingminhua",
   "copyright": "Copyright (c) 2026 LaoDing",
   "license": "MIT",