dsh-codex-subscription 1.9.0 → 1.10.0

package/AGENTS.md

@@ -6,7 +6,7 @@ Use this guide when a user asks an Agent to install, update, verify, or remove
 ## Safety
 
 - Confirm the target DSH installation and profile. Use `web` only when it is the user's target.
-- Use the exact `1.9.0` package below; do not install a moving branch.
+- Use the exact `1.10.0` package below; do not install a moving branch.
 - Never print OAuth credentials, account IDs, authorization callbacks, or credential-store contents.
 - Preserve the DSH profile, unrelated plugins, sessions, and saved sign-in.
 - Do not start, stop, or restart DSH without explicit permission.
@@ -34,13 +34,13 @@ For an official DSH installation run through npm, keep the complete `npx` prefix
 run command does not create a global `dsh` command:
 
 ```sh
-npx -y @deepseek-ai/dsh@0.1.1-rc.2 plugin --profile web add dsh-codex-subscription@1.9.0
+npx -y @deepseek-ai/dsh@0.1.1-rc.2 plugin --profile web add dsh-codex-subscription@1.10.0
 ```
 
 With an existing global `dsh` command:
 
 ```sh
-dsh plugin --profile web add dsh-codex-subscription@1.9.0
+dsh plugin --profile web add dsh-codex-subscription@1.10.0
 ```
 
 Use the same `add` command to update or repair. This is the complete package-changing operation.
@@ -58,12 +58,12 @@ dsh --profile web --dump-config
 
 For the official npm route, run the same checks with `npx -y @deepseek-ai/dsh@0.1.1-rc.2` in place of `dsh`.
 
-1. `dsh-codex-subscription` version `1.9.0` appears exactly once.
+1. `dsh-codex-subscription` version `1.10.0` appears exactly once.
 2. `codex-subscription` appears exactly once in the composed config.
 3. No unrelated plugin or profile was changed and DSH was not restarted.
 
 With permission to restart DSH, open **Settings -> Codex** and verify the settings page loads. Confirm
-that search offers both **DSH default** and **Codex subscription**, the composer quota switch defaults
+that search offers **Auto**, **DSH default**, and **Codex subscription**, the composer quota switch defaults
 to off, and `codex_image_generate` is available. Do not consume quota merely to test installation unless
 the user explicitly asks for a live model, search, or image-generation check.
 

package/README.md

@@ -65,7 +65,7 @@ Captured from the installed official DeepSeek Harness `0.1.1-rc.2` product with
 
 ## Prepare DSH
 
-This plugin is currently compatible through DeepSeek Harness `0.1.1-rc.2` and requires a ChatGPT account that currently has Codex access.
+This plugin supports the latest DeepSeek Harness release recorded in its package metadata and requires a ChatGPT account that currently has Codex access.
 
 - Do not want to configure Node.js? Use [DSH-Portable](https://github.com/WSL043/DSH-Portable), a community portable desktop distribution for Windows, macOS, and Linux.
 - Prefer the official route? Follow the [DeepSeek Harness run guide](https://github.com/deepseek-ai/deepseek-harness#run).
@@ -219,9 +219,10 @@ dsh plugin --profile web remove dsh-codex-subscription
 The ChatGPT Codex backend and DSH can change independently. This community project is not affiliated with or endorsed by DeepSeek or OpenAI.
 
 Use the [bug report form](https://github.com/WSL043/dsh-codex-subscription/issues/new?template=install-problem.yml) for project feedback.
+Use the [feature request form](https://github.com/WSL043/dsh-codex-subscription/issues/new?template=feature-request.yml) for focused product suggestions.
 For DSH plugin discussion, visit [DeepSeek Harness Discussions](https://github.com/deepseek-ai/deepseek-harness/discussions).
 Read [SECURITY.md](SECURITY.md) before reporting sensitive issues.
 
 If this project is useful, the [Star button](https://github.com/WSL043/dsh-codex-subscription/stargazers) helps more DSH users find it.
 
-[简体中文](README.md) · [MIT](LICENSE)
+[简体中文](README.zh-CN.md) · [MIT](LICENSE)

package/README.zh-CN.md

@@ -50,7 +50,7 @@ DSH-Portable 也提供相同的标准插件命令，因此同样使用上面的
 | **输入框额度** | 可选择紧凑百分比、进度条或完全关闭输入框额度显示 |
 | **安全额度重置** | 直接查看最早到期时间，并通过冷静期和知情确认主动尝试重置 |
 | **订阅搜索** | 可将全部模型的搜索明确路由到 DSH 默认搜索或已登录的 Codex 订阅 |
-| **Codex 图片生成与编辑（Beta）** | 可无参考图全新生成，也可明确选择会话图片继续编辑；支持预览、缩放和下载原图 |
+| **Codex 图片生成与编辑（Beta）** | 可无参考图全新生成，也可明确选择会话图片继续编辑；支持预览、缩放、区域备注、下载原图并衔接回当前输入框 |
 | **高速模式** | 直接在输入框切换标准或高速，无需离开当前会话 |
 | **模型感知上下文** | 可保留目录默认值、按模型启用扩展窗口，或为每个模型填写完整数字 Token 上限 |
 
@@ -66,7 +66,7 @@ DSH-Portable 也提供相同的标准插件命令，因此同样使用上面的
 
 ## 准备 DSH
 
-本插件当前适配到 DeepSeek Harness `0.1.1-rc.2`，并需要一个当前具有 Codex 使用资格的 ChatGPT 账户。
+本插件支持软件包元数据中记录的最新版 DeepSeek Harness，并需要一个当前具有 Codex 使用资格的 ChatGPT 账户。
 
 - 不想配置 Node.js：使用 [DSH-Portable](https://github.com/WSL043/DSH-Portable)。这是面向 Windows、macOS 和 Linux 的社区便携桌面分发；
 - 想按官方方式运行：查看 [DeepSeek Harness 官方说明](https://github.com/deepseek-ai/deepseek-harness#run)。
@@ -217,6 +217,7 @@ dsh plugin --profile web remove dsh-codex-subscription
 ChatGPT Codex 后端和 DSH 可能独立变化；本项目为社区项目，与 DeepSeek、OpenAI 无隶属或背书关系。
 
 本项目的问题反馈请使用[使用问题表单](https://github.com/WSL043/dsh-codex-subscription/issues/new?template=install-problem.yml)；
+明确的产品建议请使用[功能建议表单](https://github.com/WSL043/dsh-codex-subscription/issues/new?template=feature-request.yml)；
 DSH 插件交流可前往 [DeepSeek Harness Discussions](https://github.com/deepseek-ai/deepseek-harness/discussions)。
 敏感问题请先阅读 [SECURITY.md](SECURITY.md)。
 

package/lib/client.js

@@ -66,11 +66,18 @@ window.__ModuleLoader__.load({
 		const LEGACY_QUICK_QUOTA_FIELD = "quickQuotaVisible";
 		const QUICK_QUOTA_MODE_PERCENT = "percent";
 		const SEARCH_PROVIDER_FIELD = "searchProvider";
+		const SEARCH_PROVIDER_AUTO = "auto";
 		const SEARCH_PROVIDER_CODEX = "codex";
+		const DEFAULT_SEARCH_PROVIDER = SEARCH_PROVIDER_AUTO;
 		const SPEED_MODE_FIELD = "speedMode";
 		const SPEED_MODE_STANDARD = "standard";
 		const SPEED_MODE_FAST = "fast";
 		const DEFAULT_SPEED_MODE = SPEED_MODE_STANDARD;
+		const OUTPUT_VERBOSITY_FIELD = "outputVerbosity";
+		const OUTPUT_VERBOSITY_DEFAULT = "default";
+		const OUTPUT_VERBOSITY_MEDIUM = "medium";
+		const OUTPUT_VERBOSITY_HIGH = "high";
+		const DEFAULT_OUTPUT_VERBOSITY = OUTPUT_VERBOSITY_DEFAULT;
 		const CONTEXT_MODE_FIELD = "contextMode";
 		const CONTEXT_MODE_STANDARD = "standard";
 		const CONTEXT_MODE_EXTENDED = "extended";
@@ -98,7 +105,17 @@ window.__ModuleLoader__.load({
 			"gpt-5.5": 272e3,
 			"gpt-5.6": 272e3
 		});
-		const normalizeSearchProvider = (value) => ["dsh", "codex"].includes(value) ? value : "dsh";
+		const normalizeSearchProvider = (value) => [
+			"auto",
+			"dsh",
+			"codex"
+		].includes(value) ? value : DEFAULT_SEARCH_PROVIDER;
+		const normalizeOutputVerbosity = (value) => [
+			"default",
+			"low",
+			"medium",
+			"high"
+		].includes(value) ? value : DEFAULT_OUTPUT_VERBOSITY;
 		const normalizeSpeedMode = (value) => ["standard", "fast"].includes(value) ? value : DEFAULT_SPEED_MODE;
 		const normalizeContextMode = (value) => [
 			"standard",
@@ -223,7 +240,9 @@ window.__ModuleLoader__.load({
 			diagnosticsFailed: "无法生成诊断信息。",
 			feedbackOpen: "反馈问题",
 			searchTitle: "搜索来源",
-			searchScope: "此设置对所有模型和会话生效，不会随当前模型自动切换。",
+			searchScope: "自动按当前会话模型分流；手动选择会覆盖所有模型和会话。",
+			searchAuto: "自动",
+			searchAutoHint: "Codex 模型用订阅搜索，其他模型用 DSH",
 			searchDsh: "DSH 默认",
 			searchDshHint: "所有模型使用 DSH 当前搜索服务",
 			searchCodex: "Codex 订阅",
@@ -301,7 +320,16 @@ window.__ModuleLoader__.load({
 			speedStandardHint: "标准速度",
 			speedFast: "高速",
 			speedFastHint: "1.5 倍，消耗更多 Credits",
-			modelMenuAria: "模型、推理等级与速度",
+			verbosityTitle: "输出详略",
+			verbosityDefault: "模型默认",
+			verbosityDefaultHint: "使用官方模型目录推荐值",
+			verbosityLow: "简洁",
+			verbosityLowHint: "更短、更直接",
+			verbosityMedium: "均衡",
+			verbosityMediumHint: "兼顾完整性与长度",
+			verbosityHigh: "详细",
+			verbosityHighHint: "更充分的说明与结构",
+			modelMenuAria: "模型、推理等级、速度与输出详略",
 			modelLabel: "模型",
 			effortLabel: "推理等级",
 			providerDefault: "Default",
@@ -366,7 +394,9 @@ window.__ModuleLoader__.load({
 			diagnosticsFailed: "Could not create diagnostics.",
 			feedbackOpen: "Report a problem",
 			searchTitle: "Search source",
-			searchScope: "Applies to every model and session; it does not switch with the selected model.",
+			searchScope: "Auto follows the current session model; an explicit choice overrides every model and session.",
+			searchAuto: "Auto",
+			searchAutoHint: "Codex models use subscription search; other models use DSH",
 			searchDsh: "DSH default",
 			searchDshHint: "Use DSH's current search service for every model",
 			searchCodex: "Codex subscription",
@@ -444,7 +474,16 @@ window.__ModuleLoader__.load({
 			speedStandardHint: "Standard speed",
 			speedFast: "Fast",
 			speedFastHint: "1.5x; higher Credits use",
-			modelMenuAria: "Model, effort, and speed",
+			verbosityTitle: "Output detail",
+			verbosityDefault: "Model default",
+			verbosityDefaultHint: "Use the official model catalog recommendation",
+			verbosityLow: "Concise",
+			verbosityLowHint: "Shorter and more direct",
+			verbosityMedium: "Balanced",
+			verbosityMediumHint: "Balance completeness and length",
+			verbosityHigh: "Detailed",
+			verbosityHighHint: "More explanation and structure",
+			modelMenuAria: "Model, effort, speed, and output detail",
 			modelLabel: "Model",
 			effortLabel: "Effort",
 			providerDefault: "Default",
@@ -495,7 +534,7 @@ window.__ModuleLoader__.load({
 .codexSubscriptionQuotaModes{display:flex;align-items:center;gap:3px;padding:2px;border-radius:9px;background:var(--dsw-alias-bg-module-platform)}.codexSubscriptionQuotaMode{position:relative;display:flex;align-items:center;justify-content:center;min-height:26px;padding:0 9px;border-radius:7px;color:var(--dsw-alias-label-tertiary);font-size:12px;line-height:18px;cursor:pointer}.codexSubscriptionQuotaMode:has(input:checked){background:var(--dsw-alias-bg-layer-2);color:var(--dsw-alias-label-primary);box-shadow:0 0 0 1px var(--dsw-alias-border-l3)}.codexSubscriptionQuotaMode:has(input:focus-visible){outline:2px solid var(--dsw-alias-border-l3);outline-offset:1px}.codexSubscriptionQuotaMode:has(input:disabled){cursor:not-allowed;opacity:.5}.codexSubscriptionQuotaMode input{position:absolute;width:1px;height:1px;opacity:0;pointer-events:none}
 .codexSubscriptionContext{display:flex;flex-direction:column;gap:8px}.codexSubscriptionContextHead{display:flex;align-items:flex-start;justify-content:space-between;gap:12px}.codexSubscriptionContextCopy{display:flex;min-width:0;flex:1;flex-direction:column;gap:2px}.codexSubscriptionContextHint{font-size:11px;line-height:17px;color:var(--dsw-alias-label-tertiary)}.codexSubscriptionContextTrigger{height:32px;min-width:108px;display:inline-flex;align-items:center;justify-content:space-between;gap:10px;padding:0 10px 0 12px;border:0;border-radius:999px;outline:0;background:var(--dsw-alias-bg-module-platform);color:var(--dsw-alias-label-primary);font:inherit;font-size:12px;cursor:pointer}.codexSubscriptionContextTrigger:hover:not(:disabled){background:var(--dsw-alias-interactive-bg-hover)}.codexSubscriptionContextTrigger:focus-visible{box-shadow:0 0 0 2px var(--dsw-alias-border-l3)}.codexSubscriptionContextTrigger:disabled{color:var(--dsw-alias-label-dimmed);cursor:not-allowed}.codexSubscriptionContextTrigger svg{color:var(--dsw-alias-label-tertiary);transition:transform 120ms var(--ds-ease-in-out)}.codexSubscriptionContextTrigger[aria-expanded=true] svg{transform:rotate(180deg)}.codexSubscriptionContextModels{display:flex;flex-direction:column;border-top:1px solid var(--dsw-alias-border-l2)}.codexSubscriptionContextModel{min-height:42px;display:flex;align-items:center;justify-content:space-between;gap:12px;border-bottom:1px solid var(--dsw-alias-border-l2)}.codexSubscriptionContextModel:last-child{border-bottom:0}.codexSubscriptionContextModelCopy{display:flex;min-width:0;flex-direction:column}.codexSubscriptionContextModelCopy strong{font-size:12px;line-height:18px;font-weight:500}.codexSubscriptionContextModelCopy span{font-size:11px;line-height:16px;color:var(--dsw-alias-label-tertiary)}.codexSubscriptionContextInput{width:116px}
 .codexSubscriptionSwitch{position:relative;flex:0 0 auto;width:32px;height:18px;padding:0;border:1px solid var(--dsw-alias-border-l3);border-radius:999px;background:var(--dsw-alias-bg-module-platform);cursor:pointer}.codexSubscriptionSwitch:disabled{cursor:not-allowed;opacity:.5}.codexSubscriptionSwitch[aria-checked=true]{background:var(--dsw-alias-label-secondary);border-color:var(--dsw-alias-label-secondary)}.codexSubscriptionSwitchKnob{position:absolute;top:2px;left:2px;width:12px;height:12px;border-radius:50%;background:var(--dsw-alias-bg-layer-1);transition:transform 120ms var(--ds-ease-in-out)}.codexSubscriptionSwitch[aria-checked=true] .codexSubscriptionSwitchKnob{transform:translateX(14px)}
-.codexSubscriptionSearch{display:flex;flex-direction:column;gap:7px}.codexSubscriptionSearchChoices{display:grid;grid-template-columns:repeat(2,minmax(0,1fr));gap:6px}.codexSubscriptionSearchChoice{display:grid;grid-template-columns:14px minmax(0,1fr);align-items:center;column-gap:8px;min-width:0;border:1px solid var(--dsw-alias-border-l2);border-radius:10px;background:var(--dsw-alias-bg-module-platform);color:var(--dsw-alias-label-primary);padding:9px 10px;text-align:left;cursor:pointer}.codexSubscriptionSearchChoice:has(input:disabled){cursor:not-allowed;opacity:.5}.codexSubscriptionSearchChoice:has(input:checked){border-color:var(--dsw-alias-label-primary);background:var(--dsw-alias-bg-layer-2)}.codexSubscriptionSearchChoice:has(input:focus-visible){outline:2px solid var(--dsw-alias-border-l3);outline-offset:2px}.codexSubscriptionSearchInput{width:14px;height:14px;margin:0;accent-color:var(--dsw-alias-label-primary);cursor:inherit}.codexSubscriptionSearchCopy{display:block;min-width:0;pointer-events:none}.codexSubscriptionSearchCopy strong,.codexSubscriptionSearchCopy span{display:block}.codexSubscriptionSearchCopy strong{font-size:12px;line-height:18px;font-weight:500;color:var(--dsw-alias-label-secondary)}.codexSubscriptionSearchChoice:has(input:checked) strong{color:var(--dsw-alias-label-primary)}.codexSubscriptionSearchCopy span{font-size:11px;line-height:16px;color:var(--dsw-alias-label-tertiary)}.codexSubscriptionDivider{height:1px;background:var(--dsw-alias-border-l2)}
+.codexSubscriptionSearch{display:flex;flex-direction:column;gap:7px}.codexSubscriptionSearchChoices{display:grid;grid-template-columns:repeat(auto-fit,minmax(150px,1fr));gap:6px}.codexSubscriptionSearchChoice{display:grid;grid-template-columns:14px minmax(0,1fr);align-items:center;column-gap:8px;min-width:0;border:1px solid var(--dsw-alias-border-l2);border-radius:10px;background:var(--dsw-alias-bg-module-platform);color:var(--dsw-alias-label-primary);padding:9px 10px;text-align:left;cursor:pointer}.codexSubscriptionSearchChoice:has(input:disabled){cursor:not-allowed;opacity:.5}.codexSubscriptionSearchChoice:has(input:checked){border-color:var(--dsw-alias-label-primary);background:var(--dsw-alias-bg-layer-2)}.codexSubscriptionSearchChoice:has(input:focus-visible){outline:2px solid var(--dsw-alias-border-l3);outline-offset:2px}.codexSubscriptionSearchInput{width:14px;height:14px;margin:0;accent-color:var(--dsw-alias-label-primary);cursor:inherit}.codexSubscriptionSearchCopy{display:block;min-width:0;pointer-events:none}.codexSubscriptionSearchCopy strong,.codexSubscriptionSearchCopy span{display:block}.codexSubscriptionSearchCopy strong{font-size:12px;line-height:18px;font-weight:500;color:var(--dsw-alias-label-secondary)}.codexSubscriptionSearchChoice:has(input:checked) strong{color:var(--dsw-alias-label-primary)}.codexSubscriptionSearchCopy span{font-size:11px;line-height:16px;color:var(--dsw-alias-label-tertiary)}.codexSubscriptionDivider{height:1px;background:var(--dsw-alias-border-l2)}
 .codexSubscriptionAccountRow,.codexSubscriptionSectionHead{display:flex;align-items:center;justify-content:space-between;gap:12px}.codexSubscriptionStatus{display:flex;align-items:center;gap:8px;font-size:14px;line-height:22px;font-weight:500}
 .codexSubscriptionDot{width:8px;height:8px;border-radius:50%;background:var(--dsw-alias-label-dimmed)}.codexSubscriptionDot[data-state=connected]{background:var(--dsw-alias-state-success-primary)}.codexSubscriptionDot[data-state=disconnected]{background:var(--dsw-alias-state-error-primary)}
 .codexSubscriptionActions{display:flex;align-items:center;gap:8px;flex-wrap:wrap}.codexSubscriptionFlow{display:flex;flex-direction:column;gap:10px;padding:12px 14px;border-radius:10px;background:var(--dsw-alias-bg-module-platform)}
@@ -922,6 +961,7 @@ window.__ModuleLoader__.load({
 			let failedPatch;
 			let generation = 0;
 			let contextModels = [];
+			let verbosityModels = [];
 			const nativeSnapshot = () => scope.getSnapshot();
 			const read = () => {
 				const native = nativeSnapshot();
@@ -931,10 +971,12 @@ window.__ModuleLoader__.load({
 					quickQuotaMode: normalizeQuickQuotaMode(current.value?.[QUICK_QUOTA_MODE_FIELD], current.value?.[LEGACY_QUICK_QUOTA_FIELD]),
 					searchProvider: normalizeSearchProvider(current.value?.[SEARCH_PROVIDER_FIELD]),
 					speedMode: normalizeSpeedMode(current.value?.[SPEED_MODE_FIELD]),
+					outputVerbosity: normalizeOutputVerbosity(current.value?.[OUTPUT_VERBOSITY_FIELD]),
 					contextMode: normalizeContextMode(current.value?.[CONTEXT_MODE_FIELD]),
 					customContextWindow: normalizeCustomContextWindow(current.value?.[CUSTOM_CONTEXT_WINDOW_FIELD]),
 					customContextWindows: Object.fromEntries(Object.entries(CUSTOM_CONTEXT_MODEL_FIELDS).map(([modelKey, field]) => [modelKey, normalizeCustomContextWindow(current.value?.[field] ?? CUSTOM_CONTEXT_MODEL_DEFAULTS[modelKey], CUSTOM_CONTEXT_MODEL_CAPS[modelKey])])),
 					contextModels,
+					verbosityModels,
 					writable: !updating && current.status === "ready" && current.writable === true,
 					error
 				});
@@ -952,6 +994,7 @@ window.__ModuleLoader__.load({
 			});
 			const acceptFallback = (value) => {
 				contextModels = Array.isArray(value?.contextModels) ? value.contextModels : [];
+				verbosityModels = Array.isArray(value?.verbosityModels) ? value.verbosityModels : [];
 				fallbackStatus = "ready";
 				fallback = {
 					status: "ready",
@@ -959,6 +1002,7 @@ window.__ModuleLoader__.load({
 						[QUICK_QUOTA_MODE_FIELD]: normalizeQuickQuotaMode(value?.[QUICK_QUOTA_MODE_FIELD], value?.[LEGACY_QUICK_QUOTA_FIELD]),
 						[SEARCH_PROVIDER_FIELD]: normalizeSearchProvider(value?.[SEARCH_PROVIDER_FIELD]),
 						[SPEED_MODE_FIELD]: normalizeSpeedMode(value?.[SPEED_MODE_FIELD]),
+						[OUTPUT_VERBOSITY_FIELD]: normalizeOutputVerbosity(value?.[OUTPUT_VERBOSITY_FIELD]),
 						[CONTEXT_MODE_FIELD]: normalizeContextMode(value?.[CONTEXT_MODE_FIELD]),
 						[CUSTOM_CONTEXT_WINDOW_FIELD]: normalizeCustomContextWindow(value?.[CUSTOM_CONTEXT_WINDOW_FIELD]),
 						...Object.fromEntries(Object.entries(CUSTOM_CONTEXT_MODEL_FIELDS).map(([modelKey, field]) => [field, normalizeCustomContextWindow(value?.[field] ?? CUSTOM_CONTEXT_MODEL_DEFAULTS[modelKey], CUSTOM_CONTEXT_MODEL_CAPS[modelKey])]))
@@ -977,8 +1021,10 @@ window.__ModuleLoader__.load({
 				try {
 					const value = unwrap(await rpc.call(CHANNEL, "preferences/status", {}));
 					if (current !== generation) return;
-					if (nativeSnapshot().status === "ready") contextModels = Array.isArray(value?.contextModels) ? value.contextModels : [];
-					else acceptFallback(value);
+					if (nativeSnapshot().status === "ready") {
+						contextModels = Array.isArray(value?.contextModels) ? value.contextModels : [];
+						verbosityModels = Array.isArray(value?.verbosityModels) ? value.verbosityModels : [];
+					} else acceptFallback(value);
 					publish();
 				} catch {
 					if (current !== generation || nativeSnapshot().status === "ready") return;
@@ -1145,7 +1191,11 @@ window.__ModuleLoader__.load({
 					className: "codexSubscriptionSearchChoices",
 					role: "radiogroup",
 					"aria-label": t("searchTitle"),
-					children: [choice("dsh", t("searchDsh"), t("searchDshHint")), choice(SEARCH_PROVIDER_CODEX, t("searchCodex"), t("searchCodexHint"))]
+					children: [
+						choice(SEARCH_PROVIDER_AUTO, t("searchAuto"), t("searchAutoHint")),
+						choice("dsh", t("searchDsh"), t("searchDshHint")),
+						choice(SEARCH_PROVIDER_CODEX, t("searchCodex"), t("searchCodexHint"))
+					]
 				})]
 			});
 		}
@@ -1363,6 +1413,31 @@ window.__ModuleLoader__.load({
 			const speedSupported = state.current?.provider === "openai-codex" && supportsCodexFastMode(state.current?.model);
 			const speedWritable = preferenceSnapshot.status === "ready" && preferenceSnapshot.writable === true;
 			const fast = speedSupported && preferenceSnapshot.speedMode === "fast";
+			const verbositySupported = state.current?.provider === "openai-codex" && preferenceSnapshot.verbosityModels.includes(state.current?.model);
+			const verbosityWritable = preferenceSnapshot.status === "ready" && preferenceSnapshot.writable === true;
+			const verbosityItems = [
+				{
+					id: OUTPUT_VERBOSITY_DEFAULT,
+					label: t("verbosityDefault"),
+					description: t("verbosityDefaultHint")
+				},
+				{
+					id: "low",
+					label: t("verbosityLow"),
+					description: t("verbosityLowHint")
+				},
+				{
+					id: OUTPUT_VERBOSITY_MEDIUM,
+					label: t("verbosityMedium"),
+					description: t("verbosityMediumHint")
+				},
+				{
+					id: OUTPUT_VERBOSITY_HIGH,
+					label: t("verbosityHigh"),
+					description: t("verbosityHighHint")
+				}
+			];
+			const verbosityLabel = verbosityItems.find((item) => item.id === preferenceSnapshot.outputVerbosity)?.label ?? t("verbosityDefault");
 			const busy = state.status === "selecting";
 			(0, react.useEffect)(() => {
 				if (available) load();
@@ -1380,7 +1455,12 @@ window.__ModuleLoader__.load({
 			}, [open]);
 			(0, react.useEffect)(() => {
 				if (!speedSupported && pane === "speed") setPane("root");
-			}, [pane, speedSupported]);
+				if (!verbositySupported && pane === "verbosity") setPane("root");
+			}, [
+				pane,
+				speedSupported,
+				verbositySupported
+			]);
 			if (!available) return null;
 			const close = (restoreFocus = false) => {
 				setOpen(false);
@@ -1413,6 +1493,10 @@ window.__ModuleLoader__.load({
 				close(true);
 				preference.set({ [SPEED_MODE_FIELD]: speedMode });
 			};
+			const chooseVerbosity = (outputVerbosity) => {
+				close(true);
+				preference.set({ [OUTPUT_VERBOSITY_FIELD]: outputVerbosity });
+			};
 			const option = ({ key, label, description, selected, disabled, onClick }) => /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("button", {
 				type: "button",
 				role: "menuitemradio",
@@ -1545,6 +1629,19 @@ window.__ModuleLoader__.load({
 					onClick: () => chooseSpeed(SPEED_MODE_FAST)
 				})]
 			});
+			else if (pane === "verbosity") submenu = /* @__PURE__ */ (0, react_jsx_runtime.jsx)("div", {
+				className: "codexModelSelectSubmenu",
+				role: "menu",
+				"aria-label": t("verbosityTitle"),
+				children: verbosityItems.map((item) => option({
+					key: item.id,
+					label: item.label,
+					description: item.description,
+					selected: preferenceSnapshot.outputVerbosity === item.id,
+					disabled: !verbosityWritable,
+					onClick: () => chooseVerbosity(item.id)
+				}))
+			});
 			return /* @__PURE__ */ (0, react_jsx_runtime.jsxs)("div", {
 				className: "codexModelSelect",
 				ref: rootRef,
@@ -1590,6 +1687,7 @@ window.__ModuleLoader__.load({
 						cell("model", t("modelLabel"), modelLabel),
 						reasoning === void 0 ? null : cell("effort", t("effortLabel"), effortLabel),
 						speedSupported && cell("speed", t("speedTitle"), t(fast ? "speedFast" : "speedStandard")),
+						verbositySupported && cell("verbosity", t("verbosityTitle"), verbosityLabel),
 						submenu
 					]
 				}) : null]

package/lib/index.js

@@ -726,11 +726,18 @@ const QUICK_QUOTA_MODE_FIELD = "quickQuotaMode";
 const LEGACY_QUICK_QUOTA_FIELD = "quickQuotaVisible";
 const QUICK_QUOTA_MODE_PERCENT = "percent";
 const SEARCH_PROVIDER_FIELD = "searchProvider";
+const SEARCH_PROVIDER_AUTO = "auto";
 const SEARCH_PROVIDER_CODEX = "codex";
+const DEFAULT_SEARCH_PROVIDER = SEARCH_PROVIDER_AUTO;
 const SPEED_MODE_FIELD = "speedMode";
 const SPEED_MODE_STANDARD = "standard";
 const SPEED_MODE_FAST = "fast";
 const DEFAULT_SPEED_MODE = SPEED_MODE_STANDARD;
+const OUTPUT_VERBOSITY_FIELD = "outputVerbosity";
+const OUTPUT_VERBOSITY_DEFAULT = "default";
+const OUTPUT_VERBOSITY_MEDIUM = "medium";
+const OUTPUT_VERBOSITY_HIGH = "high";
+const DEFAULT_OUTPUT_VERBOSITY = OUTPUT_VERBOSITY_DEFAULT;
 const CONTEXT_MODE_FIELD = "contextMode";
 const CONTEXT_MODE_STANDARD = "standard";
 const CONTEXT_MODE_EXTENDED = "extended";
@@ -758,6 +765,12 @@ const CUSTOM_CONTEXT_MODEL_DEFAULTS = Object.freeze({
 	"gpt-5.5": 272e3,
 	"gpt-5.6": 272e3
 });
+const normalizeOutputVerbosity = (value) => [
+	"default",
+	"low",
+	"medium",
+	"high"
+].includes(value) ? value : DEFAULT_OUTPUT_VERBOSITY;
 const normalizeContextMode = (value) => [
 	"standard",
 	"extended",
@@ -829,7 +842,7 @@ const EXTENDED_CONTEXT_WINDOWS = Object.freeze({
 	"gpt-5.6-sol": 1e6,
 	"gpt-5.6-terra": 1e6
 });
-function openaiCodexSubscriptionProvider({ resolveSpeedMode = () => void 0, resolveContextMode = () => void 0, resolveCustomContextWindow = () => void 0, runNetwork = (_area, operation) => operation() } = {}) {
+function openaiCodexSubscriptionProvider({ resolveSpeedMode = () => void 0, resolveOutputVerbosity = () => OUTPUT_VERBOSITY_DEFAULT, resolveContextMode = () => void 0, resolveCustomContextWindow = () => void 0, catalog, runNetwork = (_area, operation) => operation() } = {}) {
 	const provider = createOpenAICodexProvider();
 	const requestToken = Object.freeze({
 		name: "DSH-managed Codex OAuth request token",
@@ -842,25 +855,40 @@ function openaiCodexSubscriptionProvider({ resolveSpeedMode = () => void 0, reso
 			};
 		}
 	});
-	const withSpeed = (model, options = {}) => {
-		if (resolveSpeedMode() !== "fast" || !supportsCodexFastMode(model?.id)) return options;
+	const modelMetadata = (model) => catalog?.metadata(model?.id);
+	const supportsVerbosity = (model) => modelMetadata(model)?.supportVerbosity ?? model?.id !== "gpt-5.3-codex-spark";
+	const withPreferences = (model, options = {}) => {
+		const metadata = modelMetadata(model);
+		const requestedVerbosity = resolveOutputVerbosity();
+		const textVerbosity = supportsVerbosity(model) ? requestedVerbosity === "default" ? metadata?.defaultVerbosity ?? "medium" : requestedVerbosity : void 0;
+		const fast = resolveSpeedMode() === "fast" && (metadata?.supportsFast ?? supportsCodexFastMode(model?.id));
 		const onPayload = options.onPayload;
 		return {
 			...options,
-			serviceTier: FAST_SERVICE_TIER,
+			...textVerbosity === void 0 ? {} : { textVerbosity },
+			...fast ? { serviceTier: FAST_SERVICE_TIER } : {},
 			async onPayload(payload, requestModel) {
-				const fastPayload = {
+				const preferred = {
 					...payload,
-					service_tier: FAST_SERVICE_TIER
+					...textVerbosity === void 0 ? {} : { text: {
+						...payload.text ?? {},
+						verbosity: textVerbosity
+					} },
+					...fast ? { service_tier: FAST_SERVICE_TIER } : {}
 				};
+				const next = await onPayload?.(preferred, requestModel);
 				return {
-					...await onPayload?.(fastPayload, requestModel) ?? fastPayload,
-					service_tier: FAST_SERVICE_TIER
+					...next ?? preferred,
+					...textVerbosity === void 0 ? {} : { text: {
+						...(next ?? preferred).text ?? {},
+						verbosity: textVerbosity
+					} },
+					...fast ? { service_tier: FAST_SERVICE_TIER } : {}
 				};
 			}
 		};
 	};
-	const getModels = () => provider.getModels().map((model) => {
+	const getModels = () => (catalog?.getModels(provider.getModels()) ?? provider.getModels()).map((model) => {
 		const maximum = EXTENDED_CONTEXT_WINDOWS[model.id];
 		const mode = resolveContextMode();
 		if (maximum === void 0 || !["extended", "custom"].includes(mode)) return model;
@@ -896,17 +924,147 @@ function openaiCodexSubscriptionProvider({ resolveSpeedMode = () => void 0, reso
 			apiKey: requestToken
 		}),
 		getModels,
-		stream: (model, context, options) => networkIterable(() => provider.stream(model, context, withSpeed(model, options))),
-		streamSimple: (model, context, options) => networkIterable(() => provider.streamSimple(model, context, withSpeed(model, options)))
+		stream: (model, context, options) => networkIterable(() => provider.stream(model, context, withPreferences(model, options))),
+		streamSimple: (model, context, options) => networkIterable(() => provider.streamSimple(model, context, withPreferences(model, options)))
 	});
 }
 //#endregion
 //#region src/version.js
-const PACKAGE_VERSION = "1.9.0";
+const PACKAGE_VERSION = "1.10.0";
 const USER_AGENT = `dsh-codex-subscription/${PACKAGE_VERSION}`;
 //#endregion
+//#region src/model-catalog.js
+const CODEX_MODELS_URL = `https://chatgpt.com/backend-api/codex/models?client_version=${encodeURIComponent(PACKAGE_VERSION)}`;
+const LEVELS = [
+	"off",
+	"minimal",
+	"low",
+	"medium",
+	"high",
+	"xhigh",
+	"max"
+];
+const record$4 = (value) => value !== null && typeof value === "object" && !Array.isArray(value);
+const nonEmpty$2 = (value) => typeof value === "string" && value.trim().length > 0 ? value.trim() : void 0;
+const positiveInteger = (value) => Number.isSafeInteger(value) && value > 0 ? value : void 0;
+function reasoningMap(levels) {
+	const supported = new Set((Array.isArray(levels) ? levels : []).map((level) => nonEmpty$2(record$4(level) ? level.effort : void 0)).filter(Boolean));
+	const map = Object.fromEntries(LEVELS.map((level) => [level, null]));
+	if (supported.has("none")) map.off = "none";
+	for (const level of LEVELS.slice(1)) if (supported.has(level)) map[level] = level;
+	return map;
+}
+function visibleModel(value) {
+	if (!record$4(value)) return void 0;
+	const id = nonEmpty$2(value.slug);
+	if (id === void 0 || value.supported_in_api !== true || value.visibility !== "list") return void 0;
+	const supported = Array.isArray(value.supported_reasoning_levels) ? value.supported_reasoning_levels : [];
+	const input = Array.isArray(value.input_modalities) ? value.input_modalities.filter((item) => ["text", "image"].includes(item)) : ["text", "image"];
+	return {
+		id,
+		name: nonEmpty$2(value.display_name) ?? id,
+		description: nonEmpty$2(value.description),
+		priority: Number.isFinite(value.priority) ? value.priority : 0,
+		input: input.length > 0 ? input : ["text"],
+		contextWindow: positiveInteger(value.context_window) ?? positiveInteger(value.max_context_window),
+		reasoning: supported.length > 0,
+		thinkingLevelMap: reasoningMap(supported),
+		supportVerbosity: value.support_verbosity === true,
+		defaultVerbosity: [
+			"low",
+			"medium",
+			"high"
+		].includes(value.default_verbosity) ? value.default_verbosity : void 0,
+		supportsFast: [...Array.isArray(value.additional_speed_tiers) ? value.additional_speed_tiers : [], ...Array.isArray(value.service_tiers) ? value.service_tiers.map((tier) => tier?.id) : []].some((tier) => tier === "fast" || tier === "priority")
+	};
+}
+function parseOfficialModelCatalog(value) {
+	if (!record$4(value) || !Array.isArray(value.models)) throw new Error("Codex returned a malformed model catalog");
+	const seen = /* @__PURE__ */ new Set();
+	return value.models.map(visibleModel).filter((model) => model !== void 0 && !seen.has(model.id) && seen.add(model.id)).sort((left, right) => right.priority - left.priority);
+}
+function mergeModel(baseModels, remote) {
+	const base = baseModels.find((model) => model.id === remote.id) ?? baseModels.find((model) => model.id !== "gpt-5.3-codex-spark") ?? baseModels[0];
+	if (base === void 0) return void 0;
+	return {
+		...base,
+		id: remote.id,
+		name: remote.name,
+		input: remote.input,
+		reasoning: remote.reasoning,
+		thinkingLevelMap: remote.thinkingLevelMap,
+		...remote.contextWindow === void 0 ? {} : { contextWindow: remote.contextWindow },
+		...base.id === remote.id ? {} : { cost: {
+			input: 0,
+			output: 0,
+			cacheRead: 0,
+			cacheWrite: 0
+		} }
+	};
+}
+function createOfficialModelCatalog(options = {}) {
+	const fetchCatalog = options.fetch ?? fetch;
+	let models;
+	let metadata = /* @__PURE__ */ new Map();
+	let etag;
+	let revision = 0;
+	let refreshing;
+	const refresh = async ({ signal } = {}) => {
+		if (refreshing !== void 0) return refreshing;
+		refreshing = (async () => {
+			const auth = await options.getAuth({ signal });
+			const credential = await options.readCredential({ signal });
+			const access = auth?.auth?.apiKey;
+			const accountId = credential?.type === "oauth" ? credential.accountId : void 0;
+			if (typeof access !== "string" || access.length === 0 || typeof accountId !== "string" || accountId.length === 0) return false;
+			const headers = {
+				authorization: `Bearer ${access}`,
+				"chatgpt-account-id": accountId,
+				accept: "application/json",
+				originator: "pi",
+				"user-agent": USER_AGENT,
+				...etag === void 0 ? {} : { "if-none-match": etag }
+			};
+			const response = await fetchCatalog(CODEX_MODELS_URL, {
+				method: "GET",
+				redirect: "error",
+				headers,
+				signal
+			});
+			if (response.status === 304) return false;
+			if (!response.ok) throw new Error(`Codex model catalog failed (HTTP ${response.status})`);
+			const remote = parseOfficialModelCatalog(await response.json());
+			if (remote.length === 0) throw new Error("Codex returned an empty model catalog");
+			const baseModels = options.baseModels();
+			const next = remote.map((model) => mergeModel(baseModels, model)).filter(Boolean);
+			if (next.length === 0) throw new Error("Codex model catalog has no compatible models");
+			models = next;
+			metadata = new Map(remote.map((model) => [model.id, model]));
+			etag = nonEmpty$2(response.headers.get("etag")) ?? etag;
+			revision += 1;
+			return true;
+		})().finally(() => {
+			refreshing = void 0;
+		});
+		return refreshing;
+	};
+	return Object.freeze({
+		refresh,
+		getModels: (fallback) => models ?? fallback,
+		metadata: (modelId) => metadata.get(modelId),
+		revision: () => revision,
+		clear() {
+			models = void 0;
+			metadata = /* @__PURE__ */ new Map();
+			etag = void 0;
+			revision += 1;
+		}
+	});
+}
+//#endregion
 //#region src/codex-search.js
 const CODEX_SEARCH_PROVIDER_ID = "codex-subscription";
+const CODEX_AUTO_SEARCH_PROVIDER_ID = "codex-subscription-auto";
 const CODEX_SEARCH_URL = "https://chatgpt.com/backend-api/codex/alpha/search";
 const DEFAULT_MODEL = "gpt-5.6-luna";
 const MAX_OUTPUT_TOKENS = 4096;
@@ -1021,6 +1179,19 @@ function createCodexSearchProvider(options) {
 		}
 	});
 }
+/** Route each request by its initiating model without changing the user's explicit overrides. */
+function createCodexAutoSearchProvider(options) {
+	return Object.freeze({
+		id: CODEX_AUTO_SEARCH_PROVIDER_ID,
+		available: () => true,
+		async search(request, signal) {
+			if (options.resolveModelProvider?.() === "openai-codex") return options.codex.search(request, signal);
+			const provider = options.resolveDshProvider?.();
+			if (provider === void 0 || provider.id === "codex-subscription-auto" || provider.id === "codex-subscription" || provider.available() !== true) throw new WebError("DSH default search is unavailable", "WEB_PROVIDER_UNAVAILABLE");
+			return provider.search(request, signal);
+		}
+	});
+}
 //#endregion
 //#region src/codex-images.js
 const CODEX_IMAGE_TOOL_NAME = "codex_image_generate";
@@ -1029,6 +1200,17 @@ const CODEX_IMAGE_EDIT_URL = "https://chatgpt.com/backend-api/codex/images/edits
 const IMAGE_MODEL = "gpt-image-2";
 const MAX_REFERENCE_IMAGES = 5;
 const RESPONSE_ENVELOPE_BYTES = 1024 * 1024;
+const IMAGE_QUALITIES = /* @__PURE__ */ new Set([
+	"auto",
+	"low",
+	"medium",
+	"high"
+]);
+const IMAGE_BACKGROUNDS = /* @__PURE__ */ new Set([
+	"auto",
+	"transparent",
+	"opaque"
+]);
 const PNG_SIGNATURE = Buffer.from([
 	137,
 	80,
@@ -1041,6 +1223,27 @@ const PNG_SIGNATURE = Buffer.from([
 ]);
 const record$2 = (value) => value !== null && typeof value === "object" && !Array.isArray(value);
 const nonEmpty = (value) => typeof value === "string" && value.trim().length > 0 ? value.trim() : void 0;
+function normalizeImageOptions(args) {
+	const quality = nonEmpty(args?.quality) ?? "auto";
+	const background = nonEmpty(args?.background) ?? "auto";
+	const size = nonEmpty(args?.size) ?? "auto";
+	if (!IMAGE_QUALITIES.has(quality)) throw new Error("quality must be auto, low, medium, or high");
+	if (!IMAGE_BACKGROUNDS.has(background)) throw new Error("background must be auto, transparent, or opaque");
+	if (size !== "auto") {
+		const match = /^(\d+)x(\d+)$/u.exec(size);
+		const width = Number(match?.[1]);
+		const height = Number(match?.[2]);
+		const short = Math.min(width, height);
+		const long = Math.max(width, height);
+		const pixels = width * height;
+		if (match === null || width % 16 !== 0 || height % 16 !== 0 || long > 3840 || long > short * 3 || pixels < 655360 || pixels > 8294400) throw new Error("size must be auto or a valid GPT Image 2 widthxheight resolution");
+	}
+	return {
+		quality,
+		background,
+		size
+	};
+}
 function encodedLimit(decodedBytes) {
 	return Math.ceil(decodedBytes / 3) * 4;
 }
@@ -1187,6 +1390,29 @@ function createCodexImageTool(options) {
 				required: true,
 				description: "A complete, production-ready description of the image to generate."
 			},
+			size: {
+				type: "string",
+				description: "Optional GPT Image 2 output size. Use auto unless the user requests an exact valid widthxheight resolution."
+			},
+			quality: {
+				type: "string",
+				enum: [
+					"auto",
+					"low",
+					"medium",
+					"high"
+				],
+				description: "Optional rendering quality. Use auto unless the user requests draft speed or final quality."
+			},
+			background: {
+				type: "string",
+				enum: [
+					"auto",
+					"transparent",
+					"opaque"
+				],
+				description: "Optional background mode. Request transparent only when the user needs transparency."
+			},
 			referenceImages: {
 				type: "array",
 				description: "Optional explicit references to 1-5 prior images to edit. Omit for a new image.",
@@ -1228,6 +1454,7 @@ function createCodexImageTool(options) {
 		async execute(args, exec) {
 			const prompt = nonEmpty(args.prompt);
 			if (prompt === void 0) throw new Error("prompt must be a non-empty string");
+			const imageOptions = normalizeImageOptions(args);
 			const auth = await options.getAuth({ signal: exec.signal });
 			const credential = await options.readCredential({ signal: exec.signal });
 			const access = auth?.auth?.apiKey;
@@ -1254,10 +1481,10 @@ function createCodexImageTool(options) {
 					body: JSON.stringify({
 						...images === void 0 ? {} : { images },
 						prompt,
-						background: "auto",
+						background: imageOptions.background,
 						model: IMAGE_MODEL,
-						quality: "auto",
-						size: "auto"
+						quality: imageOptions.quality,
+						size: imageOptions.size
 					}),
 					signal: exec.signal
 				});
@@ -1367,6 +1594,7 @@ async function createSubscriptionDiagnostics({ auth, preferences, login = { phas
 		configuration: {
 			contextMode: preference.contextMode,
 			quickQuotaMode: preference.quickQuotaMode,
+			...typeof preference.outputVerbosity === "string" ? { outputVerbosity: preference.outputVerbosity } : {},
 			searchProvider: preference.searchProvider,
 			speedMode: preference.speedMode,
 			writable: preference.writable === true
@@ -1783,7 +2011,7 @@ const publicError = (code, message) => ({
 		details: { issues: [] }
 	}
 });
-function createSubscriptionRpcHandler({ authHandler, usageReader, resetCreditService, preferences, diagnosticsReader }) {
+function createSubscriptionRpcHandler({ authHandler, usageReader, resetCreditService, preferences, diagnosticsReader, modelCatalog }) {
 	return async (endpoint, payload, signal) => {
 		if (endpoint === "diagnostics") try {
 			signal.throwIfAborted();
@@ -1808,13 +2036,26 @@ function createSubscriptionRpcHandler({ authHandler, usageReader, resetCreditSer
 					patch[QUICK_QUOTA_MODE_FIELD] = payload[QUICK_QUOTA_MODE_FIELD];
 				}
 				if (Object.hasOwn(payload ?? {}, "searchProvider")) {
-					if (!["dsh", "codex"].includes(payload["searchProvider"])) return publicError("internal", "Invalid search provider preference");
+					if (![
+						"auto",
+						"dsh",
+						"codex"
+					].includes(payload["searchProvider"])) return publicError("internal", "Invalid search provider preference");
 					patch[SEARCH_PROVIDER_FIELD] = payload[SEARCH_PROVIDER_FIELD];
 				}
 				if (Object.hasOwn(payload ?? {}, "speedMode")) {
 					if (!["standard", "fast"].includes(payload["speedMode"])) return publicError("internal", "Invalid speed mode preference");
 					patch[SPEED_MODE_FIELD] = payload[SPEED_MODE_FIELD];
 				}
+				if (Object.hasOwn(payload ?? {}, "outputVerbosity")) {
+					if (![
+						"default",
+						"low",
+						"medium",
+						"high"
+					].includes(payload["outputVerbosity"])) return publicError("internal", "Invalid output verbosity preference");
+					patch[OUTPUT_VERBOSITY_FIELD] = payload[OUTPUT_VERBOSITY_FIELD];
+				}
 				if (Object.hasOwn(payload ?? {}, "contextMode")) {
 					if (![
 						"standard",
@@ -1890,26 +2131,34 @@ function createSubscriptionRpcHandler({ authHandler, usageReader, resetCreditSer
 		if (endpoint === "logout" && result.ok === true) {
 			usageReader.clear();
 			resetCreditService.clear();
-		}
+			modelCatalog?.clear();
+		} else if (result.ok === true && (endpoint === "status" || result.value?.authenticated === true)) modelCatalog?.refresh({ signal: void 0 }).catch(() => {});
 		return result;
 	};
 }
 function createSearchProviderSwitcher(loader) {
 	const webEntry = () => [...loader.entries()].find((entry) => entry.options?.id === WEB_ENTRY_ID);
-	return Object.freeze({ async select(selection) {
-		const entry = webEntry();
-		const fiber = entry?.fiber;
-		if (entry === void 0 || fiber === void 0 || typeof fiber.update !== "function") throw new Error("DSH web runtime is unavailable");
-		const baseConfig = entry.options?.config ?? {};
-		const currentConfig = fiber.config ?? baseConfig;
-		const dshProvider = typeof baseConfig.searchProvider === "string" && baseConfig.searchProvider.length > 0 ? baseConfig.searchProvider : DSH_SEARCH_PROVIDER_FALLBACK;
-		const provider = selection === "codex" ? CODEX_SEARCH_PROVIDER_ID : dshProvider;
-		if (currentConfig.searchProvider === provider) return;
-		await fiber.update({
-			...currentConfig,
-			searchProvider: provider
-		}, true);
-	} });
+	const dshProviderId = () => {
+		const baseConfig = webEntry()?.options?.config ?? {};
+		return typeof baseConfig.searchProvider === "string" && baseConfig.searchProvider.length > 0 ? baseConfig.searchProvider : DSH_SEARCH_PROVIDER_FALLBACK;
+	};
+	return Object.freeze({
+		dshProviderId,
+		async select(selection) {
+			const entry = webEntry();
+			const fiber = entry?.fiber;
+			if (entry === void 0 || fiber === void 0 || typeof fiber.update !== "function") throw new Error("DSH web runtime is unavailable");
+			const baseConfig = entry.options?.config ?? {};
+			const currentConfig = fiber.config ?? baseConfig;
+			const dshProvider = dshProviderId();
+			const provider = selection === "codex" ? CODEX_SEARCH_PROVIDER_ID : selection === "auto" ? CODEX_AUTO_SEARCH_PROVIDER_ID : dshProvider;
+			if (currentConfig.searchProvider === provider) return;
+			await fiber.update({
+				...currentConfig,
+				searchProvider: provider
+			}, true);
+		}
+	});
 }
 function apply(ctx) {
 	const settings = ctx.settings.register(settingsNamespace(SETTINGS_NAMESPACE), z.object({
@@ -1919,8 +2168,18 @@ function apply(ctx) {
 			"bar"
 		]),
 		[LEGACY_QUICK_QUOTA_FIELD]: z.boolean(),
-		[SEARCH_PROVIDER_FIELD]: z.union(["dsh", SEARCH_PROVIDER_CODEX]).default("dsh"),
+		[SEARCH_PROVIDER_FIELD]: z.union([
+			SEARCH_PROVIDER_AUTO,
+			"dsh",
+			SEARCH_PROVIDER_CODEX
+		]).default(DEFAULT_SEARCH_PROVIDER),
 		[SPEED_MODE_FIELD]: z.union([SPEED_MODE_STANDARD, SPEED_MODE_FAST]).default(DEFAULT_SPEED_MODE),
+		[OUTPUT_VERBOSITY_FIELD]: z.union([
+			OUTPUT_VERBOSITY_DEFAULT,
+			"low",
+			OUTPUT_VERBOSITY_MEDIUM,
+			OUTPUT_VERBOSITY_HIGH
+		]).default(DEFAULT_OUTPUT_VERBOSITY),
 		[CONTEXT_MODE_FIELD]: z.union([
 			CONTEXT_MODE_STANDARD,
 			CONTEXT_MODE_EXTENDED,
@@ -1931,13 +2190,24 @@ function apply(ctx) {
 	}));
 	const searchProvider = createSearchProviderSwitcher(ctx.loader);
 	const network = createCodexNetworkTransport();
+	const store = new DshOAuthCredentialStore(ctx.credentials, CREDENTIAL_REF, [LEGACY_CREDENTIAL_REF]);
+	const baseProvider = createOpenAICodexProvider();
+	let resolveAuth = async () => void 0;
+	const modelCatalog = createOfficialModelCatalog({
+		getAuth: (options) => resolveAuth(options),
+		readCredential: (options) => store.read(PROVIDER, options),
+		baseModels: () => baseProvider.getModels(),
+		fetch: (input, init) => network.fetch("catalog", input, init)
+	});
 	const provider = openaiCodexSubscriptionProvider({
 		resolveSpeedMode: () => settings.get()[SPEED_MODE_FIELD],
+		resolveOutputVerbosity: () => normalizeOutputVerbosity(settings.get()[OUTPUT_VERBOSITY_FIELD]),
 		resolveContextMode: () => normalizeContextMode(settings.get()[CONTEXT_MODE_FIELD]),
 		resolveCustomContextWindow: (modelKey) => {
 			const field = CUSTOM_CONTEXT_MODEL_FIELDS[modelKey];
 			return normalizeCustomContextWindow(settings.get()[field] ?? CUSTOM_CONTEXT_MODEL_DEFAULTS[modelKey], CUSTOM_CONTEXT_MODEL_CAPS[modelKey]);
 		},
+		catalog: modelCatalog,
 		runNetwork: network.run
 	});
 	const preferences = {
@@ -1945,15 +2215,16 @@ function apply(ctx) {
 			[QUICK_QUOTA_MODE_FIELD]: normalizeQuickQuotaMode(settings.get()[QUICK_QUOTA_MODE_FIELD], settings.get()[LEGACY_QUICK_QUOTA_FIELD]),
 			[SEARCH_PROVIDER_FIELD]: settings.get()[SEARCH_PROVIDER_FIELD],
 			[SPEED_MODE_FIELD]: settings.get()[SPEED_MODE_FIELD],
+			[OUTPUT_VERBOSITY_FIELD]: normalizeOutputVerbosity(settings.get()[OUTPUT_VERBOSITY_FIELD]),
 			[CONTEXT_MODE_FIELD]: normalizeContextMode(settings.get()[CONTEXT_MODE_FIELD]),
 			[CUSTOM_CONTEXT_WINDOW_FIELD]: normalizeCustomContextWindow(settings.get()[CUSTOM_CONTEXT_WINDOW_FIELD]),
 			...Object.fromEntries(Object.entries(CUSTOM_CONTEXT_MODEL_FIELDS).map(([modelKey, field]) => [field, normalizeCustomContextWindow(settings.get()[field] ?? CUSTOM_CONTEXT_MODEL_DEFAULTS[modelKey], CUSTOM_CONTEXT_MODEL_CAPS[modelKey])])),
 			contextModels: contextModelGroups(provider.getModels()),
+			verbosityModels: provider.getModels().filter((model) => modelCatalog.metadata(model.id)?.supportVerbosity ?? model.id !== "gpt-5.3-codex-spark").map((model) => model.id),
 			writable: ctx.settings.writable
 		}),
 		update: (patch) => settings.update(patch)
 	};
-	const store = new DshOAuthCredentialStore(ctx.credentials, CREDENTIAL_REF, [LEGACY_CREDENTIAL_REF]);
 	const authModels = createModels({ credentials: store });
 	authModels.setProvider(provider);
 	const profile = Object.freeze({
@@ -1971,14 +2242,18 @@ function apply(ctx) {
 	let profileKey;
 	let profileSnapshot;
 	const profiles = () => {
-		const key = [normalizeContextMode(settings.get()[CONTEXT_MODE_FIELD]), ...Object.values(CUSTOM_CONTEXT_MODEL_FIELDS).map((field) => settings.get()[field])].join(":");
+		const key = [
+			modelCatalog.revision(),
+			normalizeContextMode(settings.get()[CONTEXT_MODE_FIELD]),
+			...Object.values(CUSTOM_CONTEXT_MODEL_FIELDS).map((field) => settings.get()[field])
+		].join(":");
 		if (key !== profileKey) {
 			profileKey = key;
 			profileSnapshot = /* @__PURE__ */ new Map([[PROVIDER, profile]]);
 		}
 		return profileSnapshot;
 	};
-	const resolveAuth = () => authModels.getAuth(PROVIDER);
+	resolveAuth = () => authModels.getAuth(PROVIDER);
 	const adapterAuth = Object.freeze({
 		credentials: store,
 		authContext: Object.freeze({
@@ -2009,7 +2284,7 @@ function apply(ctx) {
 	});
 	ctx.llm.registerAdapter([PROVIDER], adapter);
 	const currentAgent = () => ctx.get?.("agents")?.currentInitiator?.();
-	ctx.web.registerSearchProvider(createCodexSearchProvider({
+	const codexSearch = createCodexSearchProvider({
 		getAuth: resolveAuth,
 		readCredential: (options) => store.read(PROVIDER, options),
 		resolveModel: () => {
@@ -2018,6 +2293,12 @@ function apply(ctx) {
 		},
 		resolveSessionId: () => currentAgent()?.session.id,
 		fetch: (input, init) => network.fetch("search", input, init)
+	});
+	ctx.web.registerSearchProvider(codexSearch);
+	ctx.web.registerSearchProvider(createCodexAutoSearchProvider({
+		codex: codexSearch,
+		resolveModelProvider: () => currentAgent()?.session.requestContext?.()?.provider,
+		resolveDshProvider: () => ctx.web.searchProviders?.get(searchProvider.dshProviderId())
 	}));
 	ctx.effect(() => {
 		const select = async (value) => {
@@ -2053,8 +2334,12 @@ function apply(ctx) {
 			preferences,
 			login: coordinator.supportState(),
 			network
-		})
+		}),
+		modelCatalog
 	});
+	ctx.effect(() => {
+		modelCatalog.refresh().catch((error) => ctx.logger?.debug?.("could not refresh Codex model catalog: %s", error.message));
+	}, "codex-subscription: official model catalog");
 	ctx.effect(() => ctx.connection.rpc.handle(CHANNEL, handler, { authority: "loopback" }), "codex-subscription: loopback account RPC");
 }
 //#endregion

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-codex-subscription",
-  "version": "1.9.0",
+  "version": "1.10.0",
   "description": "Use ChatGPT and Codex subscriptions in DeepSeek Harness with OAuth, quota, safe resets, web search, images, and Fast mode",
   "type": "module",
   "main": "./lib/index.js",