dsh-connect-workbuddy 4.9.1 → 4.9.2

package/CHANGELOG.md

@@ -1,5 +1,21 @@
 # Changelog
 
+## 4.9.2 (2026-10-19)
+
+> 版本号是 **patch**：只改 `profiles` 里多登记一个注册表键，外加文档与注释；发往上游的请求体、路由 id、模型 id、模型选择器、目录与凭据一律未变。但**它修的是一个整区不可用的故障**——4.9.0/4.9.1 期间国际版（`workbuddy-global`）的每一个请求都在发出前失败。
+
+### Fixes
+
+- **issue #75：国际版整区报 `Unknown provider: workbuddy`，任何请求都不可能成功**
+  - **原来错在哪**：4.9.0 为修 issue #59 把两条路由的模型身份统一成 `workbuddy`（`WORKBUDDY_MODEL_IDENTITY`），理由是「只有让 `isSameModel` 判等，跨区的 thinking 才不会降级成正文」。当时的论证是：pi-ai 的 `getModel(provider, id)` 取 `providers.get(provider).getModels()`（`models.js:116–118`），**查表的键**与**模型上盖的 `provider` 字段**是分开的，所以盖上 `workbuddy` 无害。**这条论证只对按参数查表的那条路径成立，漏掉了真正跑聊天的那条**：`models.streamSimple(model, …)` 的第一帧就是 `requireProvider(model)` = `this.providers.get(model.provider)`（`models.js:351–355`），查的正是模型上盖的那个字段。而每条路由的注册表里只登记过它自己的路由 id（宿主 `dsh-llm-pi-ai` 对每个 profile 调一次 `models.setProvider(profile.piProvider)`，键取 **provider 对象的 `id`**）。于是国内版（路由 `workbuddy`、盖 `workbuddy`）自洽，国际版（路由 `workbuddy-global`、盖 `workbuddy`）在一个只有 `workbuddy-global` 的注册表里查 `workbuddy` —— 抛 `ModelsError("provider", "Unknown provider: workbuddy")`，**在请求发出之前**。
+  - **为什么测试没抓住**：#59 配的守卫测试断言的是 `getModel(provider, id)` 这种**按参数**的查表，恰好是不看盖上字段的那一种形状；于是它可以全绿，而聊天路径已经整区死掉。报错文案里也不含 "WorkBuddy" 字样，用户看到的是宿主状态一个无关的 provider 名。
+  - **现在的形状**：新增 `modelIdentityKeysOf(providerId, identity)`（`src/adapter.ts`），返回一条路由要在自己的注册表里登记的全部键。每条路由把**同一个** shim 的 provider **按两个 id 各建一个对象**登记：路由 id（`workbuddy-global`，含它自己的 shim 地址、目录、凭据）＋共享身份（`workbuddy`）。国内版路由 id 与身份相同，去重后仍只登记一个。这段构造抽成 `buildProfiles()`，初始发布与 `invalidate()`（目录刷新后重建）**共用同一份**——旧代码在 `invalidate()` 里按单键重建，刷新一次就会把国际版重新打回故障。
+  - **一个必须说破的细节**：`Models.setProvider(provider)` 存的是 `this.providers.set(provider.id, provider)`（`models.js:35–38`），键取的是**provider 对象的 `id`**，不是 `profiles` 映射的键。因此「把同一个 provider 对象发布两次」是**静默无效**的，必须给身份那个键一个 `id` 改为身份的 provider 对象。`ResolvedPiAiProviderProfile.provider` 上游注释称自己就是「`Models` 集合的键」，只有 `piProvider.id === profile.provider` 时才成立——按对象拆开后这个不变式回来了。
+  - **顺带覆盖的第二处失败点**：`Models.getAuth(providerOrModel)` 同样用 `providers.get(providerOrModel.provider)` 取 provider（`models.js:275–280`，宿主侧 `:334–341`），修之前国际版即使过了 `requireProvider` 也会以 `Provider is not configured` 倒在取凭据一步；同一个键一并覆盖。
+  - **刻意没做的**：报告者提议的一行改法（`modelIdentity = options.modelIdentity ?? providerId`）**不能采纳**——它把身份拆回按路由区分，等于回退 #59（跨区 thinking 又整批降级成正文、计费跳变重现）。也没必要把身份做成「第二条 provider 条目」：`Models.providers` 本来就是 `Map`，多一个键不影响路由集合，模型选择器、目录、凭据、`registerAdapter` 的 routes 数组全都没有变化。上游的真修法仍是让 `isSameModel` 认「provider 等价组」声明（`earendil-works/pi#6167`）。
+  - **测试**：新增 `tests/issue75-global-provider-dispatch.spec.ts`，不用 `profiles()` 的字面量断言，而是**照宿主的构造方式**建注册表（`createModels()` + 每个 profile 一次 `setProvider`）再走 `streamSimple`。其中包含：国际版真实服务的模型能在国际版注册表里被 `requireProvider` 解析且拿到**国际版自己的** `baseUrl`（不是国内版的）；国内版不被连累；**一个反证用例**手工重建 4.9.1 的单键注册表，断言它仍然失败于 `Unknown provider: workbuddy`，防止上面几条变成空转通过；`invalidate()` 之后上述性质仍在；以及 #59 的两条性质（两边盖同一个身份、跨区 thinking 仍是 `thinking`）不回退。把 `buildProfiles` 临时改回单键重跑，这组测试会以 `ModelsError: Unknown provider: workbuddy` 失败——已实测确认判别力。
+  - **文档**：`README.md` 的 issue #59 那一段原文把上述错误论证当成了依据，已改正并写明 #75 的成因与 4.9.2 的做法；`docs/audit/ISSUE59-OPTIONS.md` 的「O1′ SHIPPED」一节补了「#75 CORRECTION」，说明当时为什么论证错了、以及为什么「按你推理的那条查表路径写的测试，无法证伪从同一条路径得出的结论」。
+
 ## 4.9.1 (2026-10-19)
 
 > 版本号是 **patch**：只改账号池里两枚「当前使用」徽章的配色，外加一处 CHANGELOG 排序修正；不改请求、响应、计费、排序规则或任何配置项。发往上游的请求体与 4.9.0 逐字节相同。

package/README.md

@@ -369,7 +369,7 @@ else { await envelope.response.text() /* 或者按 SSE 消费 */ }
 - **切换区域会让上下文「变大」，甚至一换区就 `code 11115` 超限——不是网关多算你的钱**：同一份历史在两个区域的计费口径不同，根子在宿主内置的 `@earendil-works/pi-ai` 重放历史的方式。它给每条 assistant 消息算一次 `isSameModel`（`transform-messages.js:68`，判据是 provider **且** api **且** model id 三者全等）：不等的消息，其 thinking 会被**降级成普通正文**（`:88–96`，并不是丢弃），而正文是网关会数的，thinking 走的是另一个字段（字段名两边都是 `reasoning_content`）。**这一点已由 issue #59 报告者的同区增量实测证实**：`reasoning` 不计费，且同一侧自己的工具结果块按本插件实测的 455.5 tok/1k 就吃掉了它全部计费量的 97.7%。本插件把国内版与国际版注册成**两条 provider 路由**（`workbuddy` / `workbuddy-global`，见[工作原理](#工作原理)），而它们是同一个 model id——于是**换一次区，另一侧产生的全部 thinking 就整批从 thinking 桶挪进正文桶**。这就是 issue #59 里「同一份对话历史，国际网关算 1.374M、国内网关算 708K（1.94×）」的成因；反方向切回来则是同一公式的 0.696×。
   - 由此带来三个用户可见的后果：换区后 token 计数可能跳 ±30%~94%，**跳多少只取决于另一侧 thinking 的字符数，与历史总长无关**；在**同一个区域内换模型**也会触发同样的降级（`isSameModel` 的 model 那条腿），所以不换区也可能看到计数变化；跳变叠加在宿主当前的压缩阈值（`min(0.8×窗口, 窗口−128000−65536)`）上，就会让本来不超限的会话一换区立刻撞上 `code 11115 prompt is too long`。
   - **这是宿主与 pi-ai 之间的一处错配，不是本插件的 bug**：不跨 provider/模型重放 thinking 这件事本身是对的（别家模型的思考过程对当前模型没有上下文价值），错在把 `provider` 当作「是不是同一个模型」的必要条件，而这里两个 provider 字符串指的是**同一个模型的两个接入区域**。
-  - **本插件已在 4.9.0 起从插件侧绕开它**：两条路由注册的 **model 身份统一为 `workbuddy`**（`src/adapter.ts` 的 `WORKBUDDY_MODEL_IDENTITY`），也就是模型自己声明的 `provider` 字段两边相同，于是 `isSameModel` 成立、换区不再把 thinking 降级成正文。**路由 id 没有合并**——它仍是 `workbuddy` / `workbuddy-global`，各自指向自己的 shim、目录与凭据，所以一侧故障不会带走另一侧，模型选择器也仍是两组。依据是 pi-ai 自己的查表形状：`getModel(provider, id)` 取的是 `providers.get(provider).getModels()`（`models.js:116–118`），**查表用的键**与**模型上盖的 provider 字段**是分开的两件事。
+  - **本插件已在 4.9.0 起从插件侧绕开它**：两条路由注册的 **model 身份统一为 `workbuddy`**（`src/adapter.ts` 的 `WORKBUDDY_MODEL_IDENTITY`），也就是模型自己声明的 `provider` 字段两边相同，于是 `isSameModel` 成立、换区不再把 thinking 降级成正文。**路由 id 没有合并**——它仍是 `workbuddy` / `workbuddy-global`，各自指向自己的 shim、目录与凭据，所以一侧故障不会带走另一侧，模型选择器也仍是两组。**4.9.0/4.9.1 的这版修复漏了一半，国际版整区不可用（issue #75）**：上面的说法只对 `getModel(provider, id)` 这一条按**参数**查表的路径成立；真正跑聊天的 `models.streamSimple(model, …)` 第一步是 `requireProvider(model)` = `providers.get(model.provider)`（`models.js:351–355`），查的是**模型上盖的那个字段**。于是国内版（路由 `workbuddy`、盖 `workbuddy`）自洽，国际版（路由 `workbuddy-global`、盖 `workbuddy`）在只注册了 `workbuddy-global` 的注册表里查不到，**每一个请求都在发出前就抛 `Unknown provider: workbuddy`**——一次也会失败，且报错里不带任何 WorkBuddy 字样，看起来像宿主坏了。**4.9.2 起改为：每条路由把自己的注册表同时登记上「路由 id」和「共享身份」两个键**（`src/adapter.ts` 的 `modelIdentityKeysOf`），模型上盖的身份仍然两边相同（#59 不回退），而 `requireProvider` 在国际版自己的注册表里也能查到、并拿到**国际版自己的** provider 对象（含它自己的 shim 地址与凭据）。pi-ai 的注册表本来就是 `Map`（`models.js:37` 的 `providers.set(provider.id, provider)`，键取的是 **provider 对象的 `id`**，不是 profiles 映射的键），所以这是注册表原生支持的事；**路由集合、模型选择器、目录与凭据都没有变化**，多出来的那个键不出现在任何 UI 里。回归测试 `tests/issue75-global-provider-dispatch.spec.ts` 走的是宿主的真实构造（`createModels` + `setProvider` + `streamSimple`），并保留一个反证用例固定住旧的失败文案。详见 `docs/audit/ISSUE59-OPTIONS.md` 的「#75 CORRECTION」一节。
   - 升级后仍可能看到计数变化的部分：**在同一区域内换模型**仍会降级（`model` 那条腿是真的换了模型，thinking 签名未必对新模型有效，保留才是错的）；**升级之前就录在历史里的消息**带着旧的按区域 id，重放时仍按旧规则降级——这种历史只会越聊越少，不会反过来影响新录的消息。
   - 真正的修法仍在上游（让 `isSameModel` 认「provider 等价组」声明，而不是让插件去伪装自己的身份）；插件侧的兜底是 4.8.0 起的 `src/context-rescue.ts`——**一旦上游确认超限，就按块折叠历史后重发**，而不是把宿主要压缩的整段历史原样重放一遍（原样重放正是「压缩请求自己超限 → 压缩死锁」的成因）。想亲眼看到这件事发生在哪条消息上：设 `WORKBUDDY_BODY_REPORT` 抓一次 body 组成报告，它只记计数、不记正文。
 

package/lib/bin.js

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-import { C as defaultDesktopAuthDirs, F as WorkBuddyUpstreamClient, K as WORKBUDDY_APP_EXECUTABLE_ENV, O as legacyWorkbuddyOwnAuthPath, Q as findWorkbuddyAppExecutableWithSource, S as defaultDesktopAuthCandidates, a as readHostHeartbeat, c as WORKBUDDY_CONNECT_VERSION, d as FALLBACK_WORKBUDDY_MODELS_GLOBAL, j as workbuddyOwnAuthPath, l as piAiRuntimeInfo, o as workbuddyHostHeartbeatPath, ot as hostPeersInfo, r as isHeartbeatProcessAlive, u as FALLBACK_WORKBUDDY_MODELS, v as WORKBUDDY_AUTH_FILE_ENV, y as WorkBuddyCredentialStore } from "./host-heartbeat-DahwuDGO.js";
+import { C as defaultDesktopAuthDirs, F as WorkBuddyUpstreamClient, K as WORKBUDDY_APP_EXECUTABLE_ENV, O as legacyWorkbuddyOwnAuthPath, Q as findWorkbuddyAppExecutableWithSource, S as defaultDesktopAuthCandidates, a as readHostHeartbeat, c as WORKBUDDY_CONNECT_VERSION, d as FALLBACK_WORKBUDDY_MODELS_GLOBAL, j as workbuddyOwnAuthPath, l as piAiRuntimeInfo, o as workbuddyHostHeartbeatPath, ot as hostPeersInfo, r as isHeartbeatProcessAlive, u as FALLBACK_WORKBUDDY_MODELS, v as WORKBUDDY_AUTH_FILE_ENV, y as WorkBuddyCredentialStore } from "./host-heartbeat-DSMmEhuD.js";
 import { realpathSync } from "node:fs";
 import { fileURLToPath } from "node:url";
 //#region src/bin.ts

package/lib/{host-heartbeat-DahwuDGO.js → host-heartbeat-DSMmEhuD.js}

@@ -3966,7 +3966,7 @@ function generationOf(version) {
 //#endregion
 //#region src/version.ts
 /** The npm package version this build was produced from. */
-const WORKBUDDY_CONNECT_VERSION = "4.9.1";
+const WORKBUDDY_CONNECT_VERSION = "4.9.2";
 //#endregion
 //#region src/host-heartbeat.ts
 /**

package/lib/index.d.ts

@@ -1284,6 +1284,10 @@ interface WorkBuddyAdapterOptions {
    * Model identity stamped on every model this route serves; defaults to
    * {@link WORKBUDDY_MODEL_IDENTITY}. Tests override it to model a transcript
    * recorded by a host that still stamped per-region ids.
+   *
+   * It also decides the extra registry key this adapter publishes alongside
+   * its route (issue #75) — see {@link modelIdentityKeysOf}, which is what
+   * keeps an overridden stamp resolvable exactly like the production one.
    */
   modelIdentity?: string;
   /** Resolve the durable attachment service at request time, when present. */

package/lib/index.js

@@ -1,4 +1,4 @@
-import { $ as isEncryptedFieldWrapper, A as workbuddyAccountId, B as modelIdOfBody, C as defaultDesktopAuthDirs, D as isVaultAccountId, E as isEncryptedCredentialError, F as WorkBuddyUpstreamClient, G as regionOf, H as parseReasoning, I as classifyUpstreamError, J as deriveAtRestKey, K as WORKBUDDY_APP_EXECUTABLE_ENV, L as declaredTools, M as CREDENTIAL_REJECTED_CODE, N as WORKBUDDY_FALLBACK_SYSTEM_PROMPT, O as legacyWorkbuddyOwnAuthPath, P as WorkBuddyCredentialRejectedError, R as isCredentialRejectedError, S as defaultDesktopAuthCandidates, T as hasEncryptedCredentialFields, U as parseUpstreamModel, V as parseCreditMultiplier, W as prepareChatBody, X as fetchAtRestKeyPayload, Y as deriveAtRestKeyId, Z as findWorkbuddyAppExecutable, _ as WORKBUDDY_AUTH_FILENAME, a as readHostHeartbeat, at as workbuddyAppExecutableCandidates, b as WorkBuddyEncryptedCredentialError, c as WORKBUDDY_CONNECT_VERSION, ct as withFileLock, d as FALLBACK_WORKBUDDY_MODELS_GLOBAL, et as isWorkbuddyBundle, f as WorkBuddyCatalog, g as ENCRYPTED_CREDENTIAL_CODE, h as fallbackModelsFor, i as processStartTimeMs, it as readAtRestKey, j as workbuddyOwnAuthPath, k as parseWorkBuddyAuth, l as piAiRuntimeInfo, lt as writeFileAtomic, m as deriveCatalog, n as clearHostHeartbeat, nt as macosNestedAppBundles, o as workbuddyHostHeartbeatPath, p as applyContextBudgets, q as clearAtRestKeyCache, r as isHeartbeatProcessAlive, rt as openEncryptedField, s as writeHostHeartbeat, st as resolveDshHome, t as WORKBUDDY_HOST_HEARTBEAT_FILENAME, tt as macosBundleExecutable, u as FALLBACK_WORKBUDDY_MODELS, v as WORKBUDDY_AUTH_FILE_ENV, w as defaultDesktopAuthPath, x as authFileName, y as WorkBuddyCredentialStore, z as knownRegionOf } from "./host-heartbeat-DahwuDGO.js";
+import { $ as isEncryptedFieldWrapper, A as workbuddyAccountId, B as modelIdOfBody, C as defaultDesktopAuthDirs, D as isVaultAccountId, E as isEncryptedCredentialError, F as WorkBuddyUpstreamClient, G as regionOf, H as parseReasoning, I as classifyUpstreamError, J as deriveAtRestKey, K as WORKBUDDY_APP_EXECUTABLE_ENV, L as declaredTools, M as CREDENTIAL_REJECTED_CODE, N as WORKBUDDY_FALLBACK_SYSTEM_PROMPT, O as legacyWorkbuddyOwnAuthPath, P as WorkBuddyCredentialRejectedError, R as isCredentialRejectedError, S as defaultDesktopAuthCandidates, T as hasEncryptedCredentialFields, U as parseUpstreamModel, V as parseCreditMultiplier, W as prepareChatBody, X as fetchAtRestKeyPayload, Y as deriveAtRestKeyId, Z as findWorkbuddyAppExecutable, _ as WORKBUDDY_AUTH_FILENAME, a as readHostHeartbeat, at as workbuddyAppExecutableCandidates, b as WorkBuddyEncryptedCredentialError, c as WORKBUDDY_CONNECT_VERSION, ct as withFileLock, d as FALLBACK_WORKBUDDY_MODELS_GLOBAL, et as isWorkbuddyBundle, f as WorkBuddyCatalog, g as ENCRYPTED_CREDENTIAL_CODE, h as fallbackModelsFor, i as processStartTimeMs, it as readAtRestKey, j as workbuddyOwnAuthPath, k as parseWorkBuddyAuth, l as piAiRuntimeInfo, lt as writeFileAtomic, m as deriveCatalog, n as clearHostHeartbeat, nt as macosNestedAppBundles, o as workbuddyHostHeartbeatPath, p as applyContextBudgets, q as clearAtRestKeyCache, r as isHeartbeatProcessAlive, rt as openEncryptedField, s as writeHostHeartbeat, st as resolveDshHome, t as WORKBUDDY_HOST_HEARTBEAT_FILENAME, tt as macosBundleExecutable, u as FALLBACK_WORKBUDDY_MODELS, v as WORKBUDDY_AUTH_FILE_ENV, w as defaultDesktopAuthPath, x as authFileName, y as WorkBuddyCredentialStore, z as knownRegionOf } from "./host-heartbeat-DSMmEhuD.js";
 import z from "@deepseek-ai/schemastery";
 import { appendFile, mkdir, readFile, writeFile } from "node:fs/promises";
 import { randomBytes, randomUUID, timingSafeEqual } from "node:crypto";
@@ -119,25 +119,60 @@ const WORKBUDDY_PROVIDERS = {
 * byte-identical history, and a compression request that could then never fit
 * the window it was computed for.
 *
-* `provider` is also the KEY pi-ai dispatches on (`models.js:410–416`
-* `requireProvider`), so the identity cannot simply be the shared string: the
-* registry has to keep resolving per region. Both halves of that are already
-* satisfied by pi-ai's own lookup shape — `getModel(provider, id)` is
-* `getModels(provider).find(...)` (`models.js:116–118`), i.e. the KEY is the
-* route id while the STAMPED field is whatever the provider put there — so the
-* route stays `workbuddy`/`workbuddy-global` (each with its own shim, catalog
-* and credential store) while every model it serves advertises THIS id. This
-* plugin's provider registers no `fetchModels`, so nothing downstream derives
-* anything from the stamped field; the openai-completions api uses
-* `model.provider` only for provider-specific quirks (none of which match
-* these ids) and as an error-message label.
-*
-* Deliberately NOT the same string as either route: a message is only
+* The stamp is NOT inert, and reading it as inert is what broke issue #75.
+* `provider` is the KEY pi-ai dispatches on: `requireProvider` resolves
+* `this.providers.get(model.provider)` and throws
+* `Unknown provider: ${model.provider}` when it is absent (`models.js:410–416`
+* on the host's 1.1.0, `:351–357` on the 0.85.1 this repo resolves), and
+* `getAuth` resolves the same key, answering `Provider is not configured`
+* instead (`:334–341` and `:429`; `:275–280` and `:366` on 0.85.1). Both sit on
+* the LIVE chat path: `dsh-llm-pi-ai` streams through
+* `snapshot.models.streamSimple(model, …)` (`lib/index.js:1983`), which calls
+* `requireProvider` before a byte is sent (`models.js:454–459`). Each region
+* builds its OWN registry from its own adapter — `current()` registers
+* `profile.piProvider` per profile (`lib/index.js:1852–1862`) — so the
+* `workbuddy-global` region registered only `workbuddy-global` and could not
+* dispatch a model stamped `workbuddy`: every request failed with
+* `Unknown provider: workbuddy` before reaching the gateway.
+* {@link modelIdentityKeysOf} is the correction: the identity has to be a
+* registry key of its own in every region, or the stamp cannot be shared.
+*
+* Lookup by ROUTE is a separate question and stays intact: `getModel(provider,
+* id)` is `getModels(provider).find(...)` (`models.js:116–118`), i.e. it goes
+* through the map key, while the STAMPED field is whatever the provider put
+* there. So the route stays `workbuddy`/`workbuddy-global` (each with its own
+* shim, catalog and credential store) while every model it serves advertises
+* THIS id. The openai-completions api reads `model.provider` only for
+* provider-specific quirks (none of which match these ids) and as an
+* error-message label.
+*
+* Deliberately NOT the same string as the global route: a message is only
 * same-model when provider, api AND model id all match, so the shared id must
 * not be reachable as a route a user could pick — it names one model, not one
-* gateway.
+* gateway. It IS reachable as a pi-ai registry key, which is the point: those
+* are internal and never surface as a route, because `registerAdapter` takes
+* the route list explicitly (`dsh-llm/lib/index.js:1842–1861`,
+* `prepareRoutes` `:1867–1885`) and this plugin passes exactly the two region
+* ids (`src/index.ts:2069–2086`).
 */
 const WORKBUDDY_MODEL_IDENTITY = "workbuddy";
+/**
+* Registry keys an adapter serving `providerId` must publish so that every
+* stamp its models carry resolves inside its own region (issue #75).
+*
+* The route id first, then the model identity, de-duplicated — route-first so
+* `profiles()` opens with the route the adapter actually serves. CN yields one
+* key (`workbuddy` is both its route and the shared identity) and the global
+* region yields two (`workbuddy-global`, `workbuddy`); the two regions
+* therefore publish different key sets, and each must publish the shared one.
+*
+* `identity` is a parameter because {@link WorkBuddyAdapterOptions.modelIdentity}
+* can override the stamp: the keys have to follow the stamp, or a test that
+* stamps a per-region id would get the production keys back.
+*/
+function modelIdentityKeysOf(providerId, identity = WORKBUDDY_MODEL_IDENTITY) {
+	return [...new Set([providerId, identity].filter((key) => key !== void 0))];
+}
 /** Region a provider route id belongs to. */
 function regionOfProvider(provider) {
 	for (const [region, id] of Object.entries(WORKBUDDY_PROVIDERS)) if (id === provider) return region;
@@ -250,7 +285,9 @@ function workBuddyThinkingLevelMap(info) {
 *
 * `identity` — not the route id — is what `provider` carries; see
 * {@link WORKBUDDY_MODEL_IDENTITY} for why the two must differ. The route the
-* model is REACHED by is the registry key the adapter serves, not this field.
+* model is REACHED by is the registry key the adapter serves, while this
+* stamped field is what pi-ai resolves the provider BY (issue #75) — so every
+* key in {@link modelIdentityKeysOf} must be a key this route registers.
 */
 function toPiModel(info, baseUrl, identity = WORKBUDDY_MODEL_IDENTITY) {
 	const thinkingLevelMap = workBuddyThinkingLevelMap(info);
@@ -589,17 +626,21 @@ function createWorkBuddyAdapter(options) {
 		}),
 		getModels: () => buildModels()
 	};
-	const profile = {
-		provider: providerId,
+	const profileFor = (key) => ({
+		provider: key,
 		displayName: providerName,
 		streamIdleTimeoutMs: WORKBUDDY_STREAM_IDLE_TIMEOUT_MS,
 		retryPolicy: resolveRetryPolicy(void 0, "dsh-connect-workbuddy retryPolicy"),
 		configuredMaxTokens: /* @__PURE__ */ new Map(),
 		modelErrors: /* @__PURE__ */ new Map(),
 		...REQUEST_IMAGE_BUDGETS,
-		piProvider: provider
-	};
-	let profiles = /* @__PURE__ */ new Map([[providerId, profile]]);
+		piProvider: key === providerId ? provider : {
+			...provider,
+			id: key
+		}
+	});
+	const buildProfiles = () => new Map(modelIdentityKeysOf(providerId, modelIdentity).map((key) => [key, profileFor(key)]));
+	let profiles = buildProfiles();
 	return {
 		adapter: new PiAiAdapter({
 			profiles: () => profiles,
@@ -608,7 +649,7 @@ function createWorkBuddyAdapter(options) {
 			...resolveAttachments === void 0 ? {} : { resolveAttachments }
 		}),
 		invalidate: () => {
-			profiles = /* @__PURE__ */ new Map([[providerId, profile]]);
+			profiles = buildProfiles();
 		}
 	};
 }

package/package.json

@@ -2,7 +2,7 @@
   "name": "dsh-connect-workbuddy",
   "displayName": "DSH Connect WorkBuddy",
   "description": "把本机登录的 WorkBuddy 模型接入 DeepSeek Harness，并提供只读的积分概览与模型管理。",
-  "version": "4.9.1",
+  "version": "4.9.2",
   "author": "dingminhua",
   "copyright": "Copyright (c) 2026 LaoDing",
   "license": "MIT",