dsh-github-copilot 0.4.2-alpha.8 → 0.4.2-alpha.9

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/CHANGELOG.md CHANGED
@@ -1,6 +1,14 @@
1
1
  # Changelog
2
2
 
3
- ## 0.4.2-alpha.8 (prepared)
3
+ ## 0.4.2-alpha.9 (prepared)
4
+
5
+ - Require the auxiliary Auto classifier to use advertised reasoning `off` and
6
+ the isolated public adapter's native disable mode. Serialize explicit
7
+ Responses/Completions `none` or Anthropic `disabled`; fail closed when either
8
+ capability is unavailable. Preserve ordinary Chat mappings and no-tools
9
+ behavior. Synthetic wire tests do not establish supplier-side implementation.
10
+
11
+ ## 0.4.2-alpha.8 (published)
4
12
 
5
13
  - Keep every Local diagnostics population independently collapsed by default,
6
14
  including Auto. Compact summaries retain status/counts, while collection,
package/README.md CHANGED
@@ -8,7 +8,7 @@
8
8
 
9
9
  GitHub Copilot account discovery, contextual Auto routing and hosted search for DSH. The plugin reuses DSH's published native adapter, retaining a profile-wide account default with independent Session next-turn choices; it does not patch Core or maintain a second model catalog.
10
10
 
11
- **Stable release: [`0.4.1`](https://github.com/cloga/dsh-github-copilot/releases/tag/v0.4.1); prepared prerelease: `0.4.2-alpha.8`.** `0.4.2-alpha.1` was not published: its release preflight failed, and its immutable tag will not be reused. Supported host: official DSH / Windows Desktop `0.2.0-rc.2`. The stable version is also published on npm's `latest` dist-tag; prereleases remain on their channel tags. Stability-channel promotion does not expand the qualified host baseline or prove live account, model, search or Desktop health. Earlier DSH pins are historical evidence, not supported installation targets. Published, installed in a profile and loaded by a running Host are separate states.
11
+ **Stable release: [`0.4.1`](https://github.com/cloga/dsh-github-copilot/releases/tag/v0.4.1); prepared prerelease: `0.4.2-alpha.9`.** `0.4.2-alpha.1` was not published: its release preflight failed, and its immutable tag will not be reused. Supported host: official DSH / Windows Desktop `0.2.0-rc.2`. The stable version is also published on npm's `latest` dist-tag; prereleases remain on their channel tags. Stability-channel promotion does not expand the qualified host baseline or prove live account, model, search or Desktop health. Earlier DSH pins are historical evidence, not supported installation targets. Published, installed in a profile and loaded by a running Host are separate states.
12
12
 
13
13
  **Upgrade from 0.4.0:** that first stable release can interrupt concurrent managed requests with `COPILOT_PREVIEW_CREDENTIAL_CHANGED` during normal same-account native OAuth renewal. In 0.4.1, a proven plugin-owned refresh with unchanged account, endpoint and entitlements preserves already-dispatched HTTP streams; subsequent requests use freshly validated auth and metadata. Unknown/external credential notifications, revocation, cancellation and provider rejections remain explicit—not a promise of invisible recovery or automatic retry. See [the refresh boundary](./docs/copilot-accounts.md#native-oauth-refresh-and-dispatched-requests-385).
14
14
 
@@ -126,10 +126,10 @@ Supply any launcher patches with repeated `--patch /absolute/file`. Require `sup
126
126
  For a **standalone named profile**:
127
127
 
128
128
  ```sh
129
- dsh plugin --profile web add https://github.com/cloga/dsh-github-copilot/releases/download/v0.4.2-alpha.8/dsh-github-copilot-0.4.2-alpha.8.tgz
129
+ dsh plugin --profile web add https://github.com/cloga/dsh-github-copilot/releases/download/v0.4.2-alpha.9/dsh-github-copilot-0.4.2-alpha.9.tgz
130
130
  ```
131
131
 
132
- For **Desktop**, its native package manager accepts an exact npm spec, for example `dsh-github-copilot@0.4.2-alpha.8`; stable versions publish on npm's `latest` dist-tag and prereleases on their channel tags. The official rc.2 **Desktop-bundled CLI** also supports reserved-profile plugin management; a global/generic `dsh` shim is not equivalent. Qualify the installed entry before use. See [Desktop CLI qualification](./docs/npm-distribution.md#desktop-bundled-cli-on-official-rc2) and the distinct [standalone offline procedure](./docs/npm-distribution.md#controlled-offline-cli-maintenance-for-standalone-profiles). Neither path authorizes registry-policy bypass, peer patches or configuration deletion.
132
+ For **Desktop**, its native package manager accepts an exact npm spec, for example `dsh-github-copilot@0.4.2-alpha.9`; stable versions publish on npm's `latest` dist-tag and prereleases on their channel tags. The official rc.2 **Desktop-bundled CLI** also supports reserved-profile plugin management; a global/generic `dsh` shim is not equivalent. Qualify the installed entry before use. See [Desktop CLI qualification](./docs/npm-distribution.md#desktop-bundled-cli-on-official-rc2) and the distinct [standalone offline procedure](./docs/npm-distribution.md#controlled-offline-cli-maintenance-for-standalone-profiles). Neither path authorizes registry-policy bypass, peer patches or configuration deletion.
133
133
 
134
134
  After an approved reload/restart:
135
135
 
@@ -168,9 +168,9 @@ Semantic assessment is **enabled by default only for locally unknown tasks**. It
168
168
 
169
169
  Timeout/invalid output leaves demand unknown and uses the preference fallback; it never means simple or triggers another classifier. Caller cancellation and account invalidation remain terminal. Auxiliary calls add latency and supplier charges outside native answer Usage.
170
170
 
171
- The deadline is checked against a monotonic clock at preparation, native request and result boundaries, even if the event loop delays the timer. Such delays can postpone settlement; they do not authorize late dispatch or late-result acceptance. Eligible Lightweight classifiers advertising reasoning `off` take precedence, with deterministic ID tie-breaking; the request uses `off` only if the public native model also supports it. This is not a measured speed ranking. Explanations distinguish the fitting category pool and previous-model continuity from semantic merit.
171
+ The deadline is checked against a monotonic clock at preparation, native request and result boundaries, even if the event loop delays the timer. Such delays can postpone settlement; they do not authorize late dispatch or late-result acceptance. Eligible Lightweight classifiers must advertise reasoning `off`; deterministic ID tie-breaking follows. This is not a measured speed ranking. Explanations distinguish the fitting category pool and previous-model continuity from semantic merit.
172
172
 
173
- The auxiliary classifier does not inherit configured Chat reasoning effort and receives no tools. Native `off` may only omit a reasoning option, so supplier-side thinking is not guaranteed disabled. One call, bounded text input and 128 requested output tokens constrain overhead without promising a hard credit cap; ordinary Chat reasoning/tools remain unchanged.
173
+ The auxiliary classifier always requests the protocol's explicit disabled-reasoning mode (`none` for Responses/Completions, `disabled` for Anthropic) and receives no tools. A candidate is not dispatched unless both account metadata and the isolated public adapter model expose that capability; otherwise assessment fails closed. The classifier does not inherit configured Chat reasoning effort, and its off mapping is not added to shared account models. Synthetic tests verify the outgoing payload, not supplier-side implementation. One call, bounded text input and 128 requested output tokens constrain overhead without promising a hard credit cap; ordinary Chat reasoning/tools remain unchanged.
174
174
 
175
175
  The reply's selection disclosure shows captured reasons and optional content-free auxiliary milestones. `uncertain` and `insufficient-evidence` do not mean the conversation is empty. **Selection unknown** means no retained record; **Selection unavailable → Retry** means the read failed. Retry rereads that turn, not inference. Evidence is bounded and Host-local, so restart/eviction can lose it.
176
176
 
@@ -309,14 +309,14 @@ Commit attribution uses `Assisted-by` with the actual tool, never the model prov
309
309
  GitHub Releases and npm distribute the same original verified tarball. Pin a version; verify Release SHA-256 or npm `dist.integrity`. Never repack an immutable release or move/reuse its tag.
310
310
 
311
311
  ```sh
312
- curl -LO https://github.com/cloga/dsh-github-copilot/releases/download/v0.4.2-alpha.8/dsh-github-copilot-0.4.2-alpha.8.tgz
313
- curl -LO https://github.com/cloga/dsh-github-copilot/releases/download/v0.4.2-alpha.8/SHA256SUMS
312
+ curl -LO https://github.com/cloga/dsh-github-copilot/releases/download/v0.4.2-alpha.9/dsh-github-copilot-0.4.2-alpha.9.tgz
313
+ curl -LO https://github.com/cloga/dsh-github-copilot/releases/download/v0.4.2-alpha.9/SHA256SUMS
314
314
  sha256sum --check SHA256SUMS
315
315
  ```
316
316
 
317
317
  ```powershell
318
318
  $expected = (Get-Content .\SHA256SUMS).Split()[0]
319
- $actual = (Get-FileHash .\dsh-github-copilot-0.4.2-alpha.8.tgz -Algorithm SHA256).Hash.ToLowerInvariant()
319
+ $actual = (Get-FileHash .\dsh-github-copilot-0.4.2-alpha.9.tgz -Algorithm SHA256).Hash.ToLowerInvariant()
320
320
  if ($actual -cne $expected) { throw 'Release checksum mismatch' }
321
321
  ```
322
322
 
package/README.zh.md CHANGED
@@ -8,7 +8,7 @@
8
8
 
9
9
  为 DSH 提供 GitHub Copilot 账号模型发现、上下文 Auto 选模和 hosted search。插件复用 DSH 公开的原生适配器,保留 profile 全局默认账号,并支持 Session 后续 turn 指定独立账号;不修改 Core,也不维护第二套模型目录。
10
10
 
11
- **稳定版:[`0.4.1`](https://github.com/cloga/dsh-github-copilot/releases/tag/v0.4.1);待发布预览版:`0.4.2-alpha.8`。** `0.4.2-alpha.1` 的发布预检失败,因此未发布;其不可变 tag 不会复用。支持宿主:官方 DSH / Windows Desktop `0.2.0-rc.2`。稳定版也发布到 npm 的 `latest` dist-tag;预发布版本仍使用各自渠道 tag。稳定渠道不扩大已验证的宿主基线,也不证明实时账号、模型、搜索或 Desktop 健康状态。较早 DSH pin 仅是历史证据,不是当前安装目标。已发布、已安装到 profile、已被运行中的 Host 加载,是三个不同状态。
11
+ **稳定版:[`0.4.1`](https://github.com/cloga/dsh-github-copilot/releases/tag/v0.4.1);待发布预览版:`0.4.2-alpha.9`。** `0.4.2-alpha.1` 的发布预检失败,因此未发布;其不可变 tag 不会复用。支持宿主:官方 DSH / Windows Desktop `0.2.0-rc.2`。稳定版也发布到 npm 的 `latest` dist-tag;预发布版本仍使用各自渠道 tag。稳定渠道不扩大已验证的宿主基线,也不证明实时账号、模型、搜索或 Desktop 健康状态。较早 DSH pin 仅是历史证据,不是当前安装目标。已发布、已安装到 profile、已被运行中的 Host 加载,是三个不同状态。
12
12
 
13
13
  **从 0.4.0 升级:**首个正式版在同账号正常原生 OAuth 刷新时,可能以 `COPILOT_PREVIEW_CREDENTIAL_CHANGED` 中止并行托管请求。0.4.1 仅在插件拥有的原生刷新被精确验证,且账号、端点与权限连续时,保留已发出的 HTTP 流;后续请求重新验证认证与模型元数据。未知/外部凭据通知、撤销、取消与供应商拒绝仍明确报错,不承诺所有刷新无感,也不自动重试。参见[刷新边界](./docs/copilot-accounts.md#native-oauth-refresh-and-dispatched-requests-385)。
14
14
 
@@ -106,10 +106,10 @@ node package/scripts/check-search-composition.mjs --profile-dir /absolute/profil
106
106
  **独立具名 profile**:
107
107
 
108
108
  ```sh
109
- dsh plugin --profile web add https://github.com/cloga/dsh-github-copilot/releases/download/v0.4.2-alpha.8/dsh-github-copilot-0.4.2-alpha.8.tgz
109
+ dsh plugin --profile web add https://github.com/cloga/dsh-github-copilot/releases/download/v0.4.2-alpha.9/dsh-github-copilot-0.4.2-alpha.9.tgz
110
110
  ```
111
111
 
112
- **Desktop** 原生包管理器接受精确 npm spec,例如 `dsh-github-copilot@0.4.2-alpha.8`;稳定版发布到 npm `latest` dist-tag,预发布版使用对应渠道 tag。官方 rc.2 **Desktop 随附的专属 CLI**也支持管理保留 profile;全局/普通 `dsh` shim 不等价。使用前核验实际安装入口,参见[Desktop CLI 核验](./docs/npm-distribution.md#desktop-bundled-cli-on-official-rc2)及独立的[具名 profile 离线流程](./docs/npm-distribution.md#controlled-offline-cli-maintenance-for-standalone-profiles)。两种路径都不授权绕过包源政策、修改 peer 或删除配置。
112
+ **Desktop** 原生包管理器接受精确 npm spec,例如 `dsh-github-copilot@0.4.2-alpha.9`;稳定版发布到 npm `latest` dist-tag,预发布版使用对应渠道 tag。官方 rc.2 **Desktop 随附的专属 CLI**也支持管理保留 profile;全局/普通 `dsh` shim 不等价。使用前核验实际安装入口,参见[Desktop CLI 核验](./docs/npm-distribution.md#desktop-bundled-cli-on-official-rc2)及独立的[具名 profile 离线流程](./docs/npm-distribution.md#controlled-offline-cli-maintenance-for-standalone-profiles)。两种路径都不授权绕过包源政策、修改 peer 或删除配置。
113
113
 
114
114
  经批准 reload/restart 后:
115
115
 
@@ -148,9 +148,9 @@ dsh plugin --profile web add https://github.com/cloga/dsh-github-copilot/release
148
148
 
149
149
  超时/无效结果保留未知,再按偏好兜底,不解释为简单,也不换分类模型重试。用户取消和账号失效仍终止请求。辅助调用增加延迟与供应方费用,不计入原生回答 Usage。
150
150
 
151
- 准备、原生请求和结果边界均用单调时钟检查期限,不依赖定时器及时执行。事件循环延迟可能推迟结束,但不授权超期发起请求或接受结果。合格 Lightweight 分类模型优先考虑声明 reasoning `off` 的候选,同级按 ID 确定顺序;只有公开原生模型也支持时才发送 `off`。这不是实测速率排名。解释分别说明该分类的可容纳候选数和上一模型连续性,不将保留上一模型说成语义优选。
151
+ 准备、原生请求和结果边界均用单调时钟检查期限,不依赖定时器及时执行。事件循环延迟可能推迟结束,但不授权超期发起请求或接受结果。Lightweight 分类候选必须由当前账号元数据声明支持 reasoning `off`,再按确定性 ID 顺序选择。这不是实测速率排名。解释分别说明该分类的可容纳候选数和上一模型连续性,不将保留上一模型说成语义优选。
152
152
 
153
- 辅助分类器不再继承配置的 Chat reasoning effort,也不传工具。原生 `off` 可能只是省略 reasoning 参数,因此不保证供应商侧关闭思考。单次调用、有界文本输入和 128 tokens 请求输出预算限制额外开销,但不承诺额度硬上限;普通 Chat 的思考与工具保持不变。
153
+ 辅助分类器始终请求协议明确的关闭推理模式(Responses/Completions 为 `none`,Anthropic 为 `disabled`),且不传工具。只有账号元数据和隔离后的公开适配器模型都支持此能力时才发送请求;否则关闭分类请求并按不可用处理。分类器不继承配置的 Chat reasoning effort,关闭映射也不会加到共享账号模型。合成测试验证出站载荷,不证明供应商内部实现。单次调用、有界文本输入和 128 tokens 请求输出预算限制额外开销,但不承诺额度硬上限;普通 Chat 的思考与工具保持不变。
154
154
 
155
155
  回复的选模说明展示捕获的理由和可选的不含内容的辅助里程碑。`uncertain`、`insufficient-evidence` 不等于会话没有上下文。**Selection unknown**表示记录未保留,**Selection unavailable → Retry**表示读取失败;Retry 只重读该 turn,不重新推理。证据有界且仅限 Host 生命周期,重启/淘汰可能丢失。
156
156
 
@@ -289,14 +289,14 @@ Agent 从 `node scripts/agent.mjs describe --json`、`doctor --json`及`plan <ta
289
289
  GitHub Releases 和 npm 分发同一原始已校验 tarball。固定版本并核验 Release SHA-256 或 npm `dist.integrity`;不重打包不可变 Release、不移动/复用 tag。
290
290
 
291
291
  ```sh
292
- curl -LO https://github.com/cloga/dsh-github-copilot/releases/download/v0.4.2-alpha.8/dsh-github-copilot-0.4.2-alpha.8.tgz
293
- curl -LO https://github.com/cloga/dsh-github-copilot/releases/download/v0.4.2-alpha.8/SHA256SUMS
292
+ curl -LO https://github.com/cloga/dsh-github-copilot/releases/download/v0.4.2-alpha.9/dsh-github-copilot-0.4.2-alpha.9.tgz
293
+ curl -LO https://github.com/cloga/dsh-github-copilot/releases/download/v0.4.2-alpha.9/SHA256SUMS
294
294
  sha256sum --check SHA256SUMS
295
295
  ```
296
296
 
297
297
  ```powershell
298
298
  $expected = (Get-Content .\SHA256SUMS).Split()[0]
299
- $actual = (Get-FileHash .\dsh-github-copilot-0.4.2-alpha.8.tgz -Algorithm SHA256).Hash.ToLowerInvariant()
299
+ $actual = (Get-FileHash .\dsh-github-copilot-0.4.2-alpha.9.tgz -Algorithm SHA256).Hash.ToLowerInvariant()
300
300
  if ($actual -cne $expected) { throw 'Release checksum mismatch' }
301
301
  ```
302
302
 
@@ -8,7 +8,7 @@
8
8
  },
9
9
  "package": {
10
10
  "name": "dsh-github-copilot",
11
- "version": "0.4.2-alpha.8"
11
+ "version": "0.4.2-alpha.9"
12
12
  },
13
13
  "evidence": {
14
14
  "kind": "source-and-synthetic-test-inventory",
@@ -11,8 +11,8 @@
11
11
  },
12
12
  "previousReview": "Issues #332/#333 prior allocation review and post-delivery review #334; fixed-outcome diagnostics #398; user-directed budget/config trial #400.",
13
13
  "findings": "63 decisions: 44 local/unknown, 4 semantic/unknown and 15 semantic/complex, all powerful target/selection. Sol selected 57 and Astra 6; 47 marked previous-model retention, not a transition matrix. Local/unknown may include disabled/unavailable/timeout/failed/invalid-result fallbacks. Semantic/unknown may include omitted-context conservatism. Persisted aggregates discarded those diagnostics, so historical causes cannot be established.",
14
- "decision": "User explicitly approves a bounded budget/config experiment despite unproven timeout prevalence: default 30000ms, native editable integer 1000-120000ms with 8000 restoring the old budget; empty classifier selects automatically, exact account ID chooses only that eligible concrete model. Following requested overhead research, isolate classifier from configured Chat reasoning effort: reproduce inherited high in synthetic native wire, omit that auxiliary profile default and preserve ordinary Chat high. Native off/omission is not supplier disable proof; do not force low/minimal or invent a hard credit cap. Retain classification, category routing, weights, continuity, 128-token output, omitted-context safety, account/exclusion gates, cancellation and one-call/no-retry ownership. Keep fixed diagnostic cohorts and legacy unrecorded reasons. Defer benchmark/adaptive optimal-model selection until representative accuracy/latency/cost evidence and separate consent; do not infer classifier/network failure from unknown demand.",
15
- "validation": "Regression-first tests failed for the hard-coded deadline, missing configuration and inherited high-effort classifier wire. Complete local pnpm verify passed: 2735 Vitest tests with 2 expected skips, 340 tooling tests, source/test types, Agent/baseline checks, clean build and package smoke. A synthetic valid nine-second result is accepted, thirty-second deadline results rejected, explicit 8s/45s budgets reach preparation fences, and fixed/missing/excluded/disabled classifiers plus same-turn settings changes retain ownership. The native wire regression verifies omitted classifier reasoning/tools and 128 output tokens while ordinary Chat preserves high effort. Exact local pack/tarball checks passed; native SettingsForms restart/fiber fixture and required Windows/Linux results are recorded in the implementation PR when executed. Synthetic evidence does not establish live timeout prevalence, classification improvement, supplier thinking disablement or optimal model choice.",
14
+ "decision": "User explicitly approves a bounded budget/config experiment despite unproven timeout prevalence: default 30000ms, native editable integer 1000-120000ms with 8000 restoring the old budget; empty classifier selects automatically, exact account ID chooses only that eligible concrete model. Following requested overhead research, isolate the classifier from configured Chat reasoning effort and always request protocol-native reasoning disable when current account metadata and the classifier-only public model both advertise/support off: Responses and Completions none, Anthropic disabled. Fail closed before dispatch when either proof is absent; do not change shared model maps or ordinary Chat behavior. Synthetic payloads establish serialization, not supplier internals. Retain classification, category routing, weights, continuity, no-tools, 128-token output, omitted-context safety, account/exclusion gates, cancellation and one-call/no-retry ownership. Keep fixed diagnostic cohorts and legacy unrecorded reasons. Defer benchmark/adaptive optimal-model selection until representative accuracy/latency/cost evidence and separate consent; do not infer classifier/network failure from unknown demand.",
15
+ "validation": "The focused command `pnpm exec vitest run tests\\auto-task-classifier.spec.ts tests\\preview-provider.spec.ts tests\\preview-route.spec.ts --maxWorkers=2` passed (290 tests). Complete local `pnpm verify` passed: 2748 Vitest tests with 2 expected skips, 340 tooling tests, source/test types, Agent/baseline checks, clean build and package smoke. Alpha.9 exact local tarball verification passed for 206 files; its final SHA-256 is recorded in the implementation PR evidence. Synthetic wire tests verify explicit Responses/Completions `none`, Anthropic `disabled`, no classifier tools, 128 requested output tokens, fail-closed behavior without either capability, and unchanged ordinary Chat mapping. The repository preflight and this review record validator passed. Required Windows/Linux PR checks remain pending. Synthetic evidence does not establish live timeout prevalence, classification improvement, supplier-side enforcement of the disabled mode or optimal model choice.",
16
16
  "followUp": {
17
17
  "owner": "Maintainer or Agent assigned to post-delivery review issue #334",
18
18
  "trigger": "Review assessment outcomes after 200 new decisions with recorded reasons, or before the next assessment policy change. Retain the prior allocation trigger of 200 opportunities per model across 20 Sessions only if independently evidenced; aggregates cannot count Sessions.",
@@ -176,21 +176,21 @@ reviewed representative latency/accuracy/cost observations and consent, not a
176
176
  guess from model names or a timeout suspicion. Review outcomes under #334 before
177
177
  claiming the increased default improves live classification.
178
178
 
179
- Request `off` reasoning only when both current supplier metadata advertises it
180
- and the public prepared model lists an `off` effort. Supplier metadata alone
181
- does not establish native support: current managed reasoning maps can decline
182
- `off` even when advertised. The auxiliary adapter does not inherit the route's
183
- configured Chat reasoning effort: a synthetic high-effort profile previously
184
- sent `reasoning.effort: high` on the classifier request. Its isolated public
185
- profile now omits that default while ordinary Chat retains it. This prevents an
186
- explicit high-effort instruction, not supplier-side thinking.
187
-
188
- Even native `off` can mean omitting the reasoning option rather than sending a
189
- supplier disable instruction. Do not force `low` or `minimal`: on some protocols
190
- that enables thinking and enlarges the SDK's requested output budget. Retain
191
- native assembly, do not change
192
- shared model maps, guess a wire value or patch Core. No claim that reasoning is
193
- disabled for all classifiers is made.
179
+ The classifier is eligible only when current account metadata advertises
180
+ reasoning `off` and the isolated, public prepared model exposes it. It then
181
+ always requests `off`; otherwise assessment fails closed without dispatching
182
+ the classifier. The classifier-only model mapping serializes the native
183
+ disable form supported by the published adapter: Responses
184
+ `reasoning.effort: "none"`, OpenAI-compatible `reasoning_effort: "none"`, or
185
+ Anthropic `thinking.type: "disabled"`. This mapping is not added to shared
186
+ account models, so ordinary Chat behavior remains unchanged.
187
+
188
+ The auxiliary adapter also does not inherit configured Chat reasoning effort:
189
+ its isolated profile omits that default while ordinary Chat retains it. Tools
190
+ and an execution loop are absent. These checks establish the requested wire
191
+ payload in synthetic tests, not that every supplier honors the control
192
+ internally. Models lacking either advertised or native support are not used;
193
+ do not substitute `low` or `minimal`, which can enable reasoning.
194
194
 
195
195
  Tools and an execution loop are absent. Only locally unknown tasks use one
196
196
  classifier call per admitted turn; subsequent steps/retries reuse its result.
package/lib/client.js CHANGED
@@ -7223,7 +7223,7 @@ Set the \`cycles\` parameter to \`"ref"\` to resolve cyclical schemas with defs.
7223
7223
  };
7224
7224
  //#endregion
7225
7225
  //#region package.json
7226
- var version = "0.4.2-alpha.8";
7226
+ var version = "0.4.2-alpha.9";
7227
7227
  //#endregion
7228
7228
  //#region lib/types/diagnostics-client.js
7229
7229
  let owner;
package/lib/index.js CHANGED
@@ -581,7 +581,7 @@ const result$1 = strictRemoteCodec("dsh-github-copilot#DiagnosticsView", Diagnos
581
581
  var package_default = {
582
582
  name: "dsh-github-copilot",
583
583
  description: "DSH companion for GitHub Copilot sign-in, account-aware model profiles, tool compatibility, and provider-hosted search.",
584
- version: "0.4.2-alpha.8",
584
+ version: "0.4.2-alpha.9",
585
585
  releaseNotes: "CHANGELOG.md",
586
586
  publishConfig: {
587
587
  "access": "public",
@@ -4373,7 +4373,7 @@ function accountBaseURL(value) {
4373
4373
  return url.origin;
4374
4374
  }
4375
4375
  /** Materialize advertised capabilities only; model names and the local static catalog never select the API. */
4376
- function accountModelFromDescriptor(descriptor, baseURL) {
4376
+ function accountModelFromDescriptor(descriptor, baseURL, classifierReasoningOff = false) {
4377
4377
  const map = {
4378
4378
  off: null,
4379
4379
  minimal: null,
@@ -4384,9 +4384,11 @@ function accountModelFromDescriptor(descriptor, baseURL) {
4384
4384
  max: null
4385
4385
  };
4386
4386
  const advertised = new Set(descriptor.reasoning.advertisedEfforts);
4387
+ if (classifierReasoningOff && advertised.has("off")) map.off = "none";
4387
4388
  const supported = enabledLevels.filter((level) => advertised.has(level) && (descriptor.api !== "anthropic-messages" || descriptor.reasoning.adaptiveThinking === true && level !== "minimal"));
4388
4389
  for (const level of supported) map[level] = level;
4389
4390
  const mapped = new Set(supported);
4391
+ if (map.off !== null) mapped.add("off");
4390
4392
  const unmapped = [.../* @__PURE__ */ new Set([...descriptor.reasoning.unmappedEfforts, ...descriptor.reasoning.advertisedEfforts.filter((level) => !mapped.has(level))])].sort();
4391
4393
  const model = {
4392
4394
  id: descriptor.id,
@@ -4398,7 +4400,7 @@ function accountModelFromDescriptor(descriptor, baseURL) {
4398
4400
  maxTokens: descriptor.maxTokens,
4399
4401
  ...descriptor.maxInputTokens === void 0 ? {} : { maxInputTokens: descriptor.maxInputTokens },
4400
4402
  input: [...descriptor.input],
4401
- reasoning: supported.length > 0,
4403
+ reasoning: supported.length > 0 || map.off !== null,
4402
4404
  thinkingLevelMap: Object.freeze(map),
4403
4405
  unmappedReasoningEfforts: Object.freeze(unmapped),
4404
4406
  ...descriptor.reasoning.minThinkingBudget === void 0 ? {} : { minThinkingBudget: descriptor.reasoning.minThinkingBudget },
@@ -4425,11 +4427,11 @@ function accountModelFromDescriptor(descriptor, baseURL) {
4425
4427
  return Object.freeze(model);
4426
4428
  }
4427
4429
  /** Create a descriptor-driven, OAuth-only provider using native SDK serialization for all supported endpoints. */
4428
- function createAccountProvider(descriptors, guard, baseURL) {
4430
+ function createAccountProvider(descriptors, guard, baseURL, classifierReasoningOff = false) {
4429
4431
  const native = githubCopilotProvider();
4430
4432
  const oauth = native.auth.oauth;
4431
4433
  if (oauth === void 0) throw new Error("COPILOT_MANAGED_OAUTH_UNAVAILABLE");
4432
- const models = Object.freeze(descriptors.map((descriptor) => accountModelFromDescriptor(descriptor, baseURL)));
4434
+ const models = Object.freeze(descriptors.map((descriptor) => accountModelFromDescriptor(descriptor, baseURL, classifierReasoningOff)));
4433
4435
  const table = new Map(models.map((model) => [model.id, model]));
4434
4436
  if (table.size !== models.length) throw new Error("COPILOT_MANAGED_DUPLICATE_MODEL");
4435
4437
  const selected = () => {
@@ -13352,7 +13354,7 @@ function installAutoModelRouting(ctx, dependencies) {
13352
13354
  }
13353
13355
  function taskClassifierModel(models, highCostModelIds, modelId = "") {
13354
13356
  const highCost = new Set(normalizeHighCostModelIds(highCostModelIds));
13355
- return models.filter((model) => (modelId !== "" ? model.id === modelId : model.category === "lightweight") && model.input.includes("text")).toSorted((left, right) => Number(highCost.has(left.id)) - Number(highCost.has(right.id)) || Number(right.reasoning.advertisedEfforts.includes("off")) - Number(left.reasoning.advertisedEfforts.includes("off")) || left.id.localeCompare(right.id, "en")).find((model) => {
13357
+ return models.filter((model) => (modelId !== "" ? model.id === modelId : model.category === "lightweight") && model.input.includes("text") && model.reasoning.advertisedEfforts.includes("off")).toSorted((left, right) => Number(highCost.has(left.id)) - Number(highCost.has(right.id)) || Number(right.reasoning.advertisedEfforts.includes("off")) - Number(left.reasoning.advertisedEfforts.includes("off")) || left.id.localeCompare(right.id, "en")).find((model) => {
13356
13358
  const budget = calculateRequestBudget(model, 128);
13357
13359
  return budget.ok && budget.budget.hardInputLimit >= 12e3;
13358
13360
  });
@@ -13361,6 +13363,7 @@ function taskClassifierModel(models, highCostModelIds, modelId = "") {
13361
13363
  async function classifyTaskWithAdapter(model, input, signal, stream, observe, reasoningOffSupported = false, checkpoint) {
13362
13364
  checkpoint?.();
13363
13365
  if (signal.aborted) throw signal.reason;
13366
+ if (!model.reasoning.advertisedEfforts.includes("off") || !reasoningOffSupported) throw new Error("COPILOT_AUTO_CLASSIFIER_UNAVAILABLE");
13364
13367
  let output = "";
13365
13368
  let finished = false;
13366
13369
  observe?.({ stage: "adapter-started" });
@@ -13380,7 +13383,7 @@ async function classifyTaskWithAdapter(model, input, signal, stream, observe, re
13380
13383
  }]
13381
13384
  })],
13382
13385
  maxTokens: 128,
13383
- ...reasoningOffSupported && model.reasoning.advertisedEfforts.includes("off") ? { reasoningEffort: ReasoningEffortId("off") } : {}
13386
+ reasoningEffort: ReasoningEffortId("off")
13384
13387
  })) {
13385
13388
  checkpoint?.();
13386
13389
  checkpoint?.();
@@ -13726,10 +13729,11 @@ var PreviewAdapter = class extends PiAiAdapter {
13726
13729
  refreshRejected;
13727
13730
  requestBudgetSettings;
13728
13731
  accountModelSettings;
13732
+ classifierReasoningOff;
13729
13733
  replayRecovery;
13730
13734
  requestCheckpoint;
13731
13735
  captureRequest;
13732
- constructor(lifetime, optionsFor, discoverSnapshot, refreshRejected, requestBudgetSettings, accountModelSettings, replayRecovery, requestCheckpoint, captureRequest = () => void 0) {
13736
+ constructor(lifetime, optionsFor, discoverSnapshot, refreshRejected, requestBudgetSettings, accountModelSettings, classifierReasoningOff = false, replayRecovery, requestCheckpoint, captureRequest = () => void 0) {
13733
13737
  super(optionsFor());
13734
13738
  this.lifetime = lifetime;
13735
13739
  this.optionsFor = optionsFor;
@@ -13737,6 +13741,7 @@ var PreviewAdapter = class extends PiAiAdapter {
13737
13741
  this.refreshRejected = refreshRejected;
13738
13742
  this.requestBudgetSettings = requestBudgetSettings;
13739
13743
  this.accountModelSettings = accountModelSettings;
13744
+ this.classifierReasoningOff = classifierReasoningOff;
13740
13745
  this.replayRecovery = replayRecovery;
13741
13746
  this.requestCheckpoint = requestCheckpoint;
13742
13747
  this.captureRequest = captureRequest;
@@ -13891,7 +13896,7 @@ var PreviewAdapter = class extends PiAiAdapter {
13891
13896
  const signal = AbortSignal.any([lease.signal, ...options.signal === void 0 ? [] : [options.signal]]);
13892
13897
  if (signal.aborted) throw abortFailure(signal);
13893
13898
  const policy = resolveRequestBudgetPolicy(owner.requestBudgetSettings());
13894
- const model = accountModelFromDescriptor(lease.descriptor, lease.proof.baseURL);
13899
+ const model = accountModelFromDescriptor(lease.descriptor, lease.proof.baseURL, owner.classifierReasoningOff);
13895
13900
  if (options.reasoningEffort !== void 0) {
13896
13901
  const mapping = Object.entries(model.thinkingLevelMap ?? {}).find(([level]) => level === options.reasoningEffort)?.[1];
13897
13902
  if (typeof mapping !== "string" || mapping.length === 0) throw failure("COPILOT_PREVIEW_REASONING_UNSUPPORTED", "INVALID_REQUEST");
@@ -14260,7 +14265,7 @@ function createAccountRuntime(ctx, config, binding, exclusionTurns, replayRecove
14260
14265
  lastValidatedProof = void 0;
14261
14266
  }
14262
14267
  };
14263
- const { provider } = createAccountProvider(lease?.snapshot.models ?? [], guard, lease?.proof.baseURL ?? "https://api.individual.githubcopilot.com");
14268
+ const { provider } = createAccountProvider(lease?.snapshot.models ?? [], guard, lease?.proof.baseURL ?? "https://api.individual.githubcopilot.com", classifier);
14264
14269
  const profile = resolvedProfile(provider, {
14265
14270
  ...requestConfig,
14266
14271
  ...classifier ? { reasoning: void 0 } : {},
@@ -14285,7 +14290,7 @@ function createAccountRuntime(ctx, config, binding, exclusionTurns, replayRecove
14285
14290
  };
14286
14291
  };
14287
14292
  return {
14288
- adapter: new PreviewAdapter(lifetime, optionsFor, discoverSnapshot, refreshRejected, budgetSettings, cacheSettings, replayRecovery, void 0, (signal) => ctx.get("githubCopilotSessionAccounts")?.recordRequest(signal)),
14293
+ adapter: new PreviewAdapter(lifetime, optionsFor, discoverSnapshot, refreshRejected, budgetSettings, cacheSettings, false, replayRecovery, void 0, (signal) => ctx.get("githubCopilotSessionAccounts")?.recordRequest(signal)),
14289
14294
  preview: {
14290
14295
  getView,
14291
14296
  refresh,
@@ -14381,7 +14386,7 @@ function createAccountRuntime(ctx, config, binding, exclusionTurns, replayRecove
14381
14386
  });
14382
14387
  const revision = lifetime.revision;
14383
14388
  const classifierOptionsFor = (lease, hooks) => optionsFor(lease, hooks, true);
14384
- const adapter = new PreviewAdapter(lifetime, classifierOptionsFor, discoverSnapshot, refreshRejected, budgetSettings, cacheSettings, void 0, checkpoint);
14389
+ const adapter = new PreviewAdapter(lifetime, classifierOptionsFor, discoverSnapshot, refreshRejected, budgetSettings, cacheSettings, true, void 0, checkpoint);
14385
14390
  try {
14386
14391
  const prepared = await adapter.prepareCall(GITHUB_COPILOT_PREVIEW_PROVIDER_ID, model.id, signal);
14387
14392
  checkpoint?.();
@@ -68,9 +68,9 @@ export type AccountPiModel = Model<AccountModelApi> & {
68
68
  };
69
69
  export declare function copilotPublicHeaders(): Readonly<Record<string, string>>;
70
70
  /** Materialize advertised capabilities only; model names and the local static catalog never select the API. */
71
- export declare function accountModelFromDescriptor(descriptor: AccountModelDescriptor, baseURL: string): AccountPiModel;
71
+ export declare function accountModelFromDescriptor(descriptor: AccountModelDescriptor, baseURL: string, classifierReasoningOff?: boolean): AccountPiModel;
72
72
  /** Create a descriptor-driven, OAuth-only provider using native SDK serialization for all supported endpoints. */
73
- export declare function createAccountProvider(descriptors: readonly AccountModelDescriptor[], guard: AccountProviderGuard, baseURL: string): {
73
+ export declare function createAccountProvider(descriptors: readonly AccountModelDescriptor[], guard: AccountProviderGuard, baseURL: string, classifierReasoningOff?: boolean): {
74
74
  provider: CoreCompatibleProvider;
75
75
  models: readonly AccountPiModel[];
76
76
  };
package/package.json CHANGED
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "name": "dsh-github-copilot",
3
3
  "description": "DSH companion for GitHub Copilot sign-in, account-aware model profiles, tool compatibility, and provider-hosted search.",
4
- "version": "0.4.2-alpha.8",
4
+ "version": "0.4.2-alpha.9",
5
5
  "releaseNotes": "CHANGELOG.md",
6
6
  "publishConfig": {
7
7
  "access": "public",