@agent-native/core 0.94.3 → 0.95.1

package/corpus/README.md

@@ -27,5 +27,5 @@ rg -n "defineAction|useActionQuery" node_modules/@agent-native/core/corpus
 
 ## Generated Counts
 
-- core files: 2166
-- template files: 5364
+- core files: 2167
+- template files: 5375

package/corpus/core/CHANGELOG.md

@@ -1,5 +1,21 @@
 # @agent-native/core
 
+## 0.95.1
+
+### Patch Changes
+
+- 6baca78: Document Design tweak controls, inheritance, and context-specific overrides.
+- 6baca78: Show each live tool call and streamed response once, keep thinking expanded by default, and simplify tool preparation status text.
+- 6baca78: Keep public SSR responses fresh in Netlify's durable edge cache for an hour and ignore non-rendering query parameters when selecting cached responses.
+- 6baca78: Allow hosted apps to use their deployment-level Google OAuth client credentials for signed-in connection flows while keeping user OAuth tokens scoped per user.
+- 6baca78: Keep guided question forms visible while application state refreshes.
+
+## 0.95.0
+
+### Minor Changes
+
+- 2308575: Add an OpenAI-compatible PR visual recap backend for DeepSeek, Kimi, and other compatible providers.
+
 ## 0.94.3
 
 ### Patch Changes

package/corpus/core/docs/content/locales/ar-SA/template-design.mdx

@@ -364,6 +364,19 @@ pnpm dev
 طبقات DOM/الكود داخل ملف واحد معروض. تعكس لوحة الطبقات هذا التقسيم: الشاشات
 إطارات من المستوى الأعلى، وتحت الشاشة النشطة تظهر طبقات DOM/الكود المتداخلة.
 
+### Tweaks
+
+Tweaks هي عناصر تحكم مسماة لتغيير أهم الخصائص المرئية في التصميم بسرعة ومن دون
+تحرير المصدر. يمكن أن يوفر التصميم عينات ألوان وخيارات مقسمة وشرائط تمرير ومفاتيح
+لخصائص مثل لون التمييز ونصف قطر الزوايا والتباعد والكثافة والرؤية. يرتبط كل عنصر
+تحكم بخاصية CSS مخصصة يستخدمها التصميم المعروض، لذلك تتحدث المعاينة أثناء التعديل.
+
+تتبع Tweaks تسلسل المحرر. ترث نقاط التوقف المتجاوبة والحالات القيمة المحددة في
+التصميم الأساسي. يؤدي تغيير tweak أثناء تنشيط نقطة توقف أو حالة إلى إنشاء تجاوز
+لهذا السياق؛ وتؤدي إعادة ضبط عنصر التحكم إلى إزالة التجاوز واستعادة القيمة
+الموروثة. تبقى القيم المحفوظة مع التصميم. استخدم **إضافة الضوابط القرص** أو اطلب
+من الوكيل إظهار خاصية مرئية أخرى عندما لا تتضمن اللوحة عنصر التحكم المطلوب.
+
 ثبّت `/visual-edit` عندما يحتاج وكيل البرمجة إلى فتح تطبيق محلي يعمل في
 Design بدل إنشاء HTML نموذج أولي مستقل:
 

package/corpus/core/docs/content/locales/de-DE/template-design.mdx

@@ -365,6 +365,24 @@ und Bearbeitung der DOM-/Code-Ebenen in einer gerenderten Datei gedacht. Das
 Ebenenfenster spiegelt diese Trennung wider: Bildschirme sind Rahmen der
 obersten Ebene, darunter liegen die DOM-/Code-Ebenen des aktiven Bildschirms.
 
+### Tweaks
+
+Tweaks sind benannte Steuerelemente, mit denen sich die wichtigsten visuellen
+Eigenschaften eines Designs schnell und ohne Quellcodeänderung anpassen lassen.
+Ein Design kann Farbfelder, segmentierte Optionen, Schieberegler und Schalter
+für Akzentfarbe, Eckenradius, Abstände, Dichte oder Sichtbarkeit anbieten. Jedes
+Steuerelement ist mit einer vom gerenderten Design verwendeten
+CSS-Benutzereigenschaft verbunden, sodass die Vorschau beim Anpassen aktualisiert
+wird.
+
+Tweaks folgen der Kaskade des Editors. Responsive Breakpoints und Zustände erben
+den im Basisdesign festgelegten Wert. Wird ein Tweak bei aktivem Breakpoint oder
+Zustand geändert, entsteht eine Überschreibung für diesen Kontext; das Zurücksetzen
+entfernt sie und stellt den geerbten Wert wieder her. Gespeicherte Werte bleiben
+beim Design. Verwenden Sie **Fügen Sie Optimierungssteuerelemente hinzu** oder
+bitten Sie den Agenten, eine weitere visuelle Eigenschaft bereitzustellen, wenn
+das benötigte Steuerelement fehlt.
+
 Installiere `/visual-edit`, wenn ein Coding Agent eine laufende lokale App in
 Design öffnen soll, statt eigenständiges Prototype-HTML zu generieren:
 

package/corpus/core/docs/content/locales/es-ES/template-design.mdx

@@ -364,6 +364,22 @@ comportamiento del prototipo y editar las capas DOM/código dentro de un archivo
 renderizado. El panel de capas refleja esa división: las pantallas son marcos de
 nivel superior, con capas DOM/código anidadas para la pantalla activa.
 
+### Tweaks
+
+Los tweaks son controles con nombre para cambiar rápidamente las propiedades visuales
+más útiles de un diseño sin editar su código fuente. Un diseño puede ofrecer muestras
+de color, opciones segmentadas, deslizadores e interruptores para propiedades como el
+color de acento, el radio de las esquinas, el espaciado, la densidad o la visibilidad.
+Cada control está conectado a una propiedad CSS personalizada utilizada por el diseño,
+por lo que la vista previa se actualiza al ajustarlo.
+
+Los tweaks siguen la cascada del editor. Los puntos de quiebre y estados heredan el
+valor definido en el diseño base. Cambiar un tweak mientras un punto de quiebre o
+estado está activo crea una modificación para ese contexto; restablecer el control
+elimina la modificación y recupera el valor heredado. Los valores guardados permanecen
+con el diseño. Usa **Agregar controles de ajuste** o pide al agente que exponga otra
+propiedad visual cuando falte el control que necesitas.
+
 Instala `/visual-edit` cuando un agente de código deba abrir una app local en
 ejecución dentro de Design en lugar de generar HTML de prototipo independiente:
 

package/corpus/core/docs/content/locales/fr-FR/template-design.mdx

@@ -364,6 +364,23 @@ prototype et modifier les couches DOM/code dans un fichier rendu. Le panneau des
 couches reflète cette séparation : les écrans sont des cadres de niveau
 supérieur, avec les couches DOM/code imbriquées pour l'écran actif.
 
+### Tweaks
+
+Les tweaks sont des contrôles nommés qui permettent de modifier rapidement les
+propriétés visuelles les plus utiles d'un design sans éditer sa source. Un design
+peut proposer des nuanciers, des choix segmentés, des curseurs et des interrupteurs
+pour la couleur d'accent, le rayon des angles, l'espacement, la densité ou la
+visibilité. Chaque contrôle est relié à une propriété CSS personnalisée utilisée
+par le design rendu, de sorte que l'aperçu se met à jour pendant le réglage.
+
+Les tweaks suivent la cascade de l'éditeur. Les points d'arrêt responsive et les
+états héritent de la valeur définie sur le design de base. Modifier un tweak pendant
+qu'un point d'arrêt ou un état est actif crée une surcharge pour ce contexte ;
+réinitialiser le contrôle supprime la surcharge et rétablit la valeur héritée. Les
+valeurs enregistrées restent avec le design. Utilisez **Ajouter des contrôles de
+réglage** ou demandez à l'agent d'exposer une autre propriété visuelle si le contrôle
+nécessaire manque.
+
 Installez `/visual-edit` lorsqu'un agent de code doit ouvrir une app locale en
 cours d'exécution dans Design au lieu de générer du HTML de prototype autonome :
 

package/corpus/core/docs/content/locales/hi-IN/template-design.mdx

@@ -363,6 +363,22 @@ rendered file के अंदर DOM/code layers edit करने के ल
 model दिखाता है: screens top-level frames हैं, और active screen के नीचे nested
 DOM/code layers होते हैं।
 
+### Tweaks
+
+Tweaks नाम वाले controls हैं जिनसे source edit किए बिना design की सबसे उपयोगी
+visual properties जल्दी बदली जा सकती हैं। Design accent color, corner radius,
+spacing, density या visibility जैसी properties के लिए color swatches, segmented
+choices, sliders और toggles दे सकता है। हर control rendered design में उपयोग होने
+वाली CSS custom property से जुड़ा होता है, इसलिए adjustment के साथ preview update
+होता है।
+
+Tweaks editor के cascade का पालन करते हैं। Base design में set value responsive
+breakpoints और states को inherit होती है। किसी breakpoint या state के active होने
+पर tweak बदलने से उस context के लिए override बनता है; control reset करने पर override
+हट जाता है और inherited value वापस आती है। Saved values design के साथ रहते हैं।
+ज़रूरी control न हो तो **ट्विक नियंत्रण जोड़ें** का उपयोग करें या agent से कोई दूसरी
+visual property expose करने को कहें।
+
 जब coding agent को standalone prototype HTML बनाने के बजाय running local app को
 Design में खोलना हो, तब `/visual-edit` install करें:
 

package/corpus/core/docs/content/locales/ja-JP/template-design.mdx

@@ -364,6 +364,23 @@ Design は複数画面の作業では概要を最初に使います。生成後
 イヤーパネルもこの分離を反映します。画面は最上位フレームで、アクティブ
 画面の下に DOM/コードレイヤーがネストされます。
 
+### Tweaks
+
+Tweaks は、ソースを編集せずにデザインの重要な視覚プロパティをすばやく
+変更するための名前付きコントロールです。デザインは、アクセントカラー、
+角丸、間隔、密度、表示状態などについて、カラースウォッチ、セグメント選択、
+スライダー、トグルを提供できます。各コントロールはレンダリングされた
+デザインが使用する CSS カスタムプロパティに接続され、調整と同時に
+プレビューが更新されます。
+
+Tweaks はエディターのカスケードに従います。ベースデザインで設定した値は、
+レスポンシブブレークポイントと状態に継承されます。ブレークポイントまたは
+状態がアクティブなときに tweak を変更すると、そのコンテキスト用の
+オーバーライドが作成されます。リセットするとオーバーライドが削除され、
+継承値に戻ります。保存した値はデザインとともに保持されます。必要な
+コントロールがない場合は、**微調整コントロールを追加する**を使用するか、
+別の視覚プロパティを公開するようエージェントに依頼してください。
+
 コーディングエージェントがスタンドアロンのプロトタイプ HTML を生成する
 代わりに、実行中のローカルアプリを Design で開く必要がある場合は
 `/visual-edit` をインストールします。

package/corpus/core/docs/content/locales/ko-KR/template-design.mdx

@@ -363,6 +363,21 @@ Single-screen mode는 스크롤, prototype behavior와의 상호작용, 하나
 반영합니다. screens는 top-level frames이고, active screen 아래에 nested
 DOM/code layers가 표시됩니다.
 
+### Tweaks
+
+Tweaks는 소스를 편집하지 않고 디자인에서 가장 유용한 시각 속성을 빠르게
+바꾸는 이름 있는 컨트롤입니다. 디자인은 강조색, 모서리 반경, 간격, 밀도,
+표시 여부 등에 대해 색상 견본, 분할 선택, 슬라이더, 토글을 제공할 수 있습니다.
+각 컨트롤은 렌더링된 디자인이 사용하는 CSS 사용자 지정 속성에 연결되므로
+조정하는 즉시 미리보기가 업데이트됩니다.
+
+Tweaks는 편집기의 계단식 규칙을 따릅니다. 기본 디자인에서 설정한 값은 반응형
+중단점과 상태에 상속됩니다. 중단점이나 상태가 활성화된 동안 tweak를 변경하면
+해당 컨텍스트에 대한 재정의가 만들어지고, 컨트롤을 재설정하면 재정의가
+제거되어 상속된 값으로 돌아갑니다. 저장된 값은 디자인과 함께 유지됩니다.
+필요한 컨트롤이 없다면 **조정 컨트롤 추가**를 사용하거나 에이전트에게 다른
+시각 속성을 노출해 달라고 요청하세요.
+
 코딩 에이전트가 독립형 prototype HTML을 생성하는 대신 실행 중인 로컬 앱을
 Design에서 열어야 할 때 `/visual-edit`를 설치하세요:
 

package/corpus/core/docs/content/locales/pt-BR/template-design.mdx

@@ -364,6 +364,22 @@ protótipo e editar as camadas DOM/código dentro de um arquivo renderizado. O
 painel de camadas reflete essa divisão: telas são frames de nível superior, com
 camadas DOM/código aninhadas para a tela ativa.
 
+### Tweaks
+
+Tweaks são controles nomeados para alterar rapidamente as propriedades visuais
+mais úteis de um design sem editar o código-fonte. Um design pode oferecer
+amostras de cor, opções segmentadas, controles deslizantes e alternâncias para
+propriedades como cor de destaque, raio dos cantos, espaçamento, densidade ou
+visibilidade. Cada controle se conecta a uma propriedade CSS personalizada usada
+pelo design renderizado, então a visualização é atualizada durante o ajuste.
+
+Tweaks seguem a cascata do editor. Breakpoints responsivos e estados herdam o
+valor definido no design base. Alterar um tweak enquanto um breakpoint ou estado
+está ativo cria uma substituição para esse contexto; redefinir o controle remove
+a substituição e restaura o valor herdado. Os valores salvos permanecem com o
+design. Use **Adicionar controles de ajuste** ou peça ao agente para expor outra
+propriedade visual quando o controle necessário não estiver no painel.
+
 Instale `/visual-edit` quando um agente de código precisar abrir um app local em
 execução no Design em vez de gerar HTML de protótipo independente:
 

package/corpus/core/docs/content/locales/zh-CN/template-design.mdx

@@ -360,6 +360,17 @@ Design 对多屏工作采用概览优先。生成或进行大范围更新后，
 单屏模式用于滚动、与原型行为交互，以及编辑一个已渲染文件内的 DOM/代码
 层。图层面板也反映这种分工：屏幕是顶层框架，活动屏幕下嵌套 DOM/代码层。
 
+### Tweaks
+
+Tweaks 是具名控件，无需编辑源代码即可快速调整设计中最常用的视觉属性。设计
+可以为强调色、圆角、间距、密度或可见性等属性提供色板、分段选项、滑块和开关。
+每个控件都连接到渲染设计使用的 CSS 自定义属性，因此调整时预览会立即更新。
+
+Tweaks 遵循编辑器的级联规则。基础设计中设置的值会由响应式断点和状态继承。
+在某个断点或状态激活时更改 tweak，会为该上下文创建覆盖；重置控件会移除覆盖
+并恢复继承值。保存的 tweak 值会随设计保留。如果面板没有所需控件，请使用
+**添加调整控件**，或让代理公开另一个视觉属性。
+
 当编码代理需要在 Design 中打开正在运行的本地应用，而不是生成独立原型 HTML
 时，请安装 `/visual-edit`：
 

package/corpus/core/docs/content/locales/zh-TW/template-design.mdx

@@ -361,6 +361,18 @@ Design 對多畫面工作採用概覽優先。產生或進行大範圍更新後
 式碼圖層。圖層面板也反映這種分工：畫面是頂層框架，作用中畫面下嵌套
 DOM/程式碼圖層。
 
+### Tweaks
+
+Tweaks 是具名控制項，不必編輯原始碼即可快速調整設計中最實用的視覺屬性。
+設計可以為強調色、圓角、間距、密度或可見性等屬性提供色票、分段選項、滑桿
+和開關。每個控制項都連接到渲染設計使用的 CSS 自訂屬性，因此調整時預覽會
+立即更新。
+
+Tweaks 遵循編輯器的級聯規則。基礎設計中設定的值會由響應式中斷點和狀態繼
+承。在某個中斷點或狀態作用時變更 tweak，會為該情境建立覆寫；重設控制項會
+移除覆寫並恢復繼承值。儲存的 tweak 值會隨設計保留。如果面板沒有需要的控制
+項，請使用 **新增調整控制項**，或要求代理公開另一個視覺屬性。
+
 當編碼代理需要在 Design 中開啟正在執行的本機應用，而不是產生獨立原型 HTML
 時，請安裝 `/visual-edit`：
 

package/corpus/core/docs/content/pr-visual-recap.mdx

@@ -7,7 +7,7 @@ description: "A GitHub Action that runs your repo's visual-recap skill on every
 
 PR Visual Recap is a GitHub Action that turns every pull request into a **visual code review**. On each push, an LLM coding agent runs the latest bundled [`visual-recap`](/docs/template-plan) skill (or your repo's committed copy when `VISUAL_RECAP_SKILL_SOURCE=repo`) against the PR diff, publishes a structured recap plan to the hosted Plans app, shows an informational `Visual Recap` check while it runs, and upserts **one sticky PR comment** that links to the interactive plan with an **inline screenshot** embedded right in the comment.
 
-This is not a deterministic diff renderer. The action invokes a real coding agent (Claude Code CLI by default, or OpenAI Codex CLI) that reads the change, decides what matters, and authors the recap by calling the Plans MCP tool `create-visual-recap` — the same tool the `/visual-recap` slash command uses. You get a high-altitude, schema/API/before-after view of the change instead of a wall of raw diff.
+This is not a deterministic diff renderer. The action invokes a real coding agent (Claude Code CLI by default, OpenAI Codex CLI, or Agent-Native Code with an OpenAI-compatible provider) that reads the change, decides what matters, and authors the recap by calling the Plans MCP tool `create-visual-recap` — the same tool the `/visual-recap` slash command uses. You get a high-altitude, schema/API/before-after view of the change instead of a wall of raw diff.
 
 The recap is **informational and non-blocking**. It creates a check row so reviewers can see that generation is in progress, but it is not a required check, it never blocks the PR, and it never replaces reading the actual diff. The sticky comment is a review aid, not a sign-off.
 
@@ -36,7 +36,7 @@ On each PR push, the workflow:
   <div class="diagram-arrow diagram-muted" aria-hidden="true">&rarr;</div>
   <div class="diagram-node">
     Coding agent<br /><small class="diagram-muted"
-      >Claude Code / Codex reads diff</small
+      >configured backend reads diff</small
     >
   </div>
   <div class="diagram-arrow diagram-muted" aria-hidden="true">&rarr;</div>
@@ -127,28 +127,32 @@ pins its committed `visual-recap` folder, set the repository variable
 
 Choose which coding agent runs the skill with the `VISUAL_RECAP_AGENT` repository variable:
 
-| `VISUAL_RECAP_AGENT` | Coding agent     | Required API key    |
-| -------------------- | ---------------- | ------------------- |
-| `claude` _(default)_ | Claude Code CLI  | `ANTHROPIC_API_KEY` |
-| `codex`              | OpenAI Codex CLI | `OPENAI_API_KEY`    |
+| `VISUAL_RECAP_AGENT` | Coding agent                            | Required API key       | Required variables                            |
+| -------------------- | --------------------------------------- | ---------------------- | --------------------------------------------- |
+| `claude` _(default)_ | Claude Code CLI                         | `ANTHROPIC_API_KEY`    | —                                             |
+| `codex`              | OpenAI Codex CLI                        | `OPENAI_API_KEY`       | —                                             |
+| `openai-compatible`  | Agent-Native Code + compatible provider | `VISUAL_RECAP_API_KEY` | `VISUAL_RECAP_BASE_URL`, `VISUAL_RECAP_MODEL` |
 
 If the variable is unset, the action uses `claude`.
 
 ## Model and reasoning
 
-Beyond the backend, two repository variables tune _how_ the agent runs:
+Beyond the backend, these repository variables tune how the agent runs:
 
-- **`VISUAL_RECAP_MODEL`** pins the model passed to the CLI (`--model`) — for example `gpt-5.6-sol` for Codex, or a Claude model id. Leave it unset to use the CLI's own default model.
+- **`VISUAL_RECAP_MODEL`** pins the model passed to the CLI (`--model`) — for example `gpt-5.6-sol` for Codex, a Claude model id, or `deepseek-chat` for DeepSeek. It is required for `openai-compatible`; Claude and Codex may leave it unset to use their CLI default.
 - **`VISUAL_RECAP_REASONING`** sets the reasoning depth: `none`, `minimal`, `low`, `medium`, `high`, or `xhigh`. It applies to the Codex backend; Claude's reasoning is model-driven, so this variable is ignored there.
+- **`VISUAL_RECAP_BASE_URL`** selects the HTTP(S) base URL for the `openai-compatible` backend. Use the provider's OpenAI-compatible endpoint, such as DeepSeek or Kimi, without embedding credentials in the URL.
 - **`VISUAL_RECAP_SKILL_SOURCE`** controls prompt freshness: `auto`/unset uses the latest bundled skill guidance, while `repo` pins to the committed repo-local `visual-recap` skill folder.
 
 For example, to run the recap on Codex with GPT-5.6 Sol at high reasoning, set the repository variables `VISUAL_RECAP_AGENT=codex`, `VISUAL_RECAP_MODEL=gpt-5.6-sol`, and `VISUAL_RECAP_REASONING=high`.
 
+For DeepSeek, set `VISUAL_RECAP_AGENT=openai-compatible`, `VISUAL_RECAP_BASE_URL=https://api.deepseek.com/v1`, `VISUAL_RECAP_MODEL=deepseek-chat`, and the `VISUAL_RECAP_API_KEY` secret. Kimi works the same way with its OpenAI-compatible base URL and model id. Other OpenAI-compatible providers use the same three settings.
+
 ## Secrets and variables
 
 Set these in your repository's **Settings → Secrets and variables → Actions**.
 
-### Secrets (only two required)
+### Secrets for the default backend
 
 | Secret              | Purpose                                                                                                                            |
 | ------------------- | ---------------------------------------------------------------------------------------------------------------------------------- |
@@ -190,8 +194,10 @@ real token.
 | Secret / variable        | Default                         | When you need it                                                                                                                                |
 | ------------------------ | ------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------- |
 | `OPENAI_API_KEY`         | —                               | Secret. Set together with `VISUAL_RECAP_AGENT=codex` to run the recap with Codex instead.                                                       |
-| `VISUAL_RECAP_AGENT`     | `claude`                        | Variable. Selects the coding-agent backend (`claude` or `codex`).                                                                               |
-| `VISUAL_RECAP_MODEL`     | each CLI's default              | Variable. Pins the model — e.g. `gpt-5.6-sol` for Codex, or a Claude model id. Unset uses the CLI's own default.                                |
+| `VISUAL_RECAP_API_KEY`   | —                               | Secret. Set with `VISUAL_RECAP_AGENT=openai-compatible` for DeepSeek, Kimi, or another OpenAI-compatible provider.                              |
+| `VISUAL_RECAP_AGENT`     | `claude`                        | Variable. Selects the coding-agent backend (`claude`, `codex`, or `openai-compatible`).                                                         |
+| `VISUAL_RECAP_BASE_URL`  | required for compatible backend | Variable. HTTP(S) base URL for the OpenAI-compatible backend; credentials are rejected.                                                         |
+| `VISUAL_RECAP_MODEL`     | required for compatible backend | Variable. Provider model id for OpenAI-compatible backends; optional for Claude/Codex, where unset uses the CLI default.                        |
 | `VISUAL_RECAP_REASONING` | each model's default            | Variable. Reasoning depth: `none`, `minimal`, `low`, `medium`, `high`, or `xhigh`. Applies to the Codex backend.                                |
 | `RECAP_CLI_VERSION`      | `latest`                        | Variable. Pins the `@agent-native/core` CLI version the workflow installs — e.g. `1.5.0`. See [Version pinning](#version-pinning-copy-variant). |
 | `PLAN_RECAP_APP_URL`     | `https://plan.agent-native.com` | Secret. Only when self-hosting the Plans app at a different origin.                                                                             |
@@ -214,7 +220,7 @@ This also means you can merge the workflow file **before** the secrets exist: wi
 
 If you want to generate recaps for fork PRs, a second workflow file is available: `.github/workflows/pr-visual-recap-fork.yml`. It uses `pull_request_target` (which runs with base-repo secrets) but never checks out or executes fork code. Trusted fork authors with GitHub author association `OWNER`, `MEMBER`, or `COLLABORATOR` run automatically. Outside fork PRs require an explicit **per-head maintainer opt-in** via a fresh `recap` label event before the recap agent runs.
 
-To install it, copy the file from [BuilderIO/agent-native](https://github.com/BuilderIO/agent-native/blob/main/.github/workflows/pr-visual-recap-fork.yml) into your repo's `.github/workflows/` directory alongside the existing `pr-visual-recap.yml`. The same secrets (`PLAN_RECAP_TOKEN`, `ANTHROPIC_API_KEY`) apply.
+To install it, copy the file from [BuilderIO/agent-native](https://github.com/BuilderIO/agent-native/blob/main/.github/workflows/pr-visual-recap-fork.yml) into your repo's `.github/workflows/` directory alongside the existing `pr-visual-recap.yml`. The same backend configuration applies, including `VISUAL_RECAP_API_KEY`, `VISUAL_RECAP_BASE_URL`, and `VISUAL_RECAP_MODEL` when using an OpenAI-compatible provider.
 
 <Diagram id="doc-block-h3lne0" title="Fork PR consent gate" summary={"Fork PRs get no secrets by default; trusted authors run automatically, and outside contributors require a fresh maintainer recap label."}>
 
@@ -454,10 +460,12 @@ jobs:
       PLAN_RECAP_TOKEN: ${{ secrets.PLAN_RECAP_TOKEN }}
       ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
       OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
+      VISUAL_RECAP_API_KEY: ${{ secrets.VISUAL_RECAP_API_KEY }}
       PLAN_RECAP_APP_URL: ${{ secrets.PLAN_RECAP_APP_URL }}
     with:
       agent: ${{ vars.VISUAL_RECAP_AGENT || 'claude' }}
       model: ${{ vars.VISUAL_RECAP_MODEL || '' }}
+      base-url: ${{ vars.VISUAL_RECAP_BASE_URL || '' }}
       reasoning: ${{ vars.VISUAL_RECAP_REASONING || '' }}
       skill-source: ${{ vars.VISUAL_RECAP_SKILL_SOURCE || 'auto' }}
       # cli-version: "latest"  # pin to a specific @agent-native/core version

package/corpus/core/docs/content/template-design.mdx

@@ -367,6 +367,22 @@ editing the DOM/code layers inside one rendered file. The layers panel reflects
 that split: screens are top-level frames, with nested DOM/code layers for the
 active screen.
 
+### Tweaks
+
+Tweaks are named controls for quickly changing the most useful visual properties
+of a design without editing its source. A design can expose color swatches,
+segmented choices, sliders, and toggles for properties such as accent color,
+corner radius, spacing, density, or visibility. Each control is connected to a
+CSS custom property used by the rendered design, so the preview updates as you
+adjust it.
+
+Tweaks follow the editor's cascade. A value set on the base design is inherited
+by its responsive breakpoints and states. Changing a tweak while a breakpoint or
+state is active creates an override for that context; resetting the control
+removes the override and returns to the inherited value. Saved tweak values stay
+with the design. Use **Add tweak controls** or ask the agent to expose another
+visual property when the panel does not include the control you need.
+
 Install `/visual-edit` when a coding agent should open a running local app in
 Design instead of generating standalone prototype HTML:
 

package/corpus/core/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@agent-native/core",
-  "version": "0.94.3",
+  "version": "0.95.1",
   "description": "Framework for agent-native application development — where AI agents and UI share SQL state, actions, and context",
   "homepage": "https://github.com/BuilderIO/agent-native#readme",
   "bugs": {
@@ -346,7 +346,7 @@
     "@tabler/icons-react": "^3.40.0",
     "@tailwindcss/typography": "^0.5.20",
     "@tailwindcss/vite": "catalog:",
-    "@tanstack/react-query": "^5.99.2",
+    "@tanstack/react-query": "^5.101.2",
     "@types/better-sqlite3": "^7.6.13",
     "@types/diff-match-patch": "^1.0.36",
     "@types/express": "^5.0.6",

package/corpus/core/src/agent/engine/ai-sdk-engine.ts

@@ -205,6 +205,7 @@ class AISDKEngine implements AgentEngine {
   readonly label: string;
   readonly defaultModel: string;
   readonly supportedModels: readonly string[];
+  readonly preserveCustomModels: boolean;
   readonly capabilities: EngineCapabilities;
 
   private readonly provider: AISDKProvider;
@@ -219,6 +220,8 @@ class AISDKEngine implements AgentEngine {
     this.label = `${capitalize(provider)} (AI SDK)`;
     this.defaultModel = config.model ?? PROVIDER_DEFAULT_MODELS[provider];
     this.supportedModels = PROVIDER_SUPPORTED_MODELS[provider];
+    this.preserveCustomModels =
+      provider === "openai" && Boolean(config.baseUrl);
     this.capabilities = PROVIDER_CAPABILITIES[provider];
     this.apiKey =
       config.apiKey ??

package/corpus/core/src/agent/engine/registry.ts

@@ -180,12 +180,17 @@ function findLatestSupportedVersionMatch(
 }
 
 export function normalizeModelForEngine(
-  engine: Pick<AgentEngine, "name" | "defaultModel" | "supportedModels">,
+  engine: Pick<
+    AgentEngine,
+    "name" | "defaultModel" | "supportedModels" | "preserveCustomModels"
+  >,
   model: string | null | undefined,
 ): string {
   const candidate = typeof model === "string" ? model.trim() : "";
   if (!candidate) return engine.defaultModel;
 
+  if (engine.preserveCustomModels) return candidate;
+
   if (engine.supportedModels.length === 0) return candidate;
 
   if (candidate === "auto" || engine.supportedModels.includes(candidate)) {

package/corpus/core/src/agent/engine/types.ts

@@ -297,6 +297,8 @@ export interface AgentEngine {
   readonly defaultModel: string;
   /** Models this engine supports */
   readonly supportedModels: readonly string[];
+  /** Whether the configured endpoint accepts provider-defined model ids. */
+  readonly preserveCustomModels?: boolean;
   /** Capability flags used to gate provider-specific features */
   readonly capabilities: EngineCapabilities;
 

package/corpus/core/src/agent/production-agent.ts

@@ -3776,11 +3776,13 @@ export async function runAgentLoop(opts: {
         const result = sourceSweepGuard.message;
         send({
           type: "tool_start",
+          id: toolCall.id,
           tool: toolCall.name,
           input: toolCall.input as Record<string, string>,
         });
         send({
           type: "tool_done",
+          id: toolCall.id,
           tool: toolCall.name,
           input: toolCall.input as Record<string, unknown>,
           result,
@@ -3800,11 +3802,13 @@ export async function runAgentLoop(opts: {
         const result = sourceSweepDelegationGuard;
         send({
           type: "tool_start",
+          id: toolCall.id,
           tool: toolCall.name,
           input: toolCall.input as Record<string, string>,
         });
         send({
           type: "tool_done",
+          id: toolCall.id,
           tool: toolCall.name,
           input: toolCall.input as Record<string, unknown>,
           result,
@@ -3826,11 +3830,13 @@ export async function runAgentLoop(opts: {
         );
         send({
           type: "tool_start",
+          id: toolCall.id,
           tool: toolCall.name,
           input: toolCall.input as Record<string, string>,
         });
         send({
           type: "tool_done",
+          id: toolCall.id,
           tool: toolCall.name,
           input: toolCall.input as Record<string, unknown>,
           result,
@@ -3875,6 +3881,7 @@ export async function runAgentLoop(opts: {
         if (mustApprove) {
           send({
             type: "tool_start",
+            id: toolCall.id,
             tool: toolCall.name,
             input: toolCall.input as Record<string, string>,
           });
@@ -3913,6 +3920,7 @@ export async function runAgentLoop(opts: {
             `can run. The turn is paused; do not retry.`;
           send({
             type: "tool_done",
+            id: toolCall.id,
             tool: toolCall.name,
             input: toolCall.input as Record<string, unknown>,
             result,
@@ -3973,11 +3981,13 @@ export async function runAgentLoop(opts: {
             recordedResult;
           send({
             type: "tool_start",
+            id: toolCall.id,
             tool: toolCall.name,
             input: toolCall.input as Record<string, string>,
           });
           send({
             type: "tool_done",
+            id: toolCall.id,
             tool: toolCall.name,
             input: toolCall.input as Record<string, unknown>,
             result,
@@ -4027,11 +4037,13 @@ export async function runAgentLoop(opts: {
               ledgerResult;
             send({
               type: "tool_start",
+              id: toolCall.id,
               tool: toolCall.name,
               input: toolCall.input as Record<string, string>,
             });
             send({
               type: "tool_done",
+              id: toolCall.id,
               tool: toolCall.name,
               input: toolCall.input as Record<string, unknown>,
               result,
@@ -4056,11 +4068,13 @@ export async function runAgentLoop(opts: {
             `or split the request into smaller pieces.`;
           send({
             type: "tool_start",
+            id: toolCall.id,
             tool: toolCall.name,
             input: toolCall.input as Record<string, string>,
           });
           send({
             type: "tool_done",
+            id: toolCall.id,
             tool: toolCall.name,
             input: toolCall.input as Record<string, unknown>,
             result,
@@ -4107,6 +4121,7 @@ export async function runAgentLoop(opts: {
 
       send({
         type: "tool_start",
+        id: toolCall.id,
         tool: toolCall.name,
         input: toolCall.input as Record<string, string>,
       });
@@ -4122,6 +4137,7 @@ export async function runAgentLoop(opts: {
         );
         send({
           type: "tool_done",
+          id: toolCall.id,
           tool: toolCall.name,
           input: toolCall.input as Record<string, unknown>,
           result,
@@ -4153,6 +4169,7 @@ export async function runAgentLoop(opts: {
         );
         send({
           type: "tool_done",
+          id: toolCall.id,
           tool: toolCall.name,
           input: toolCall.input as Record<string, unknown>,
           result,
@@ -4184,6 +4201,7 @@ export async function runAgentLoop(opts: {
           `Previous result:\n${previousResult}`;
         send({
           type: "tool_done",
+          id: toolCall.id,
           tool: toolCall.name,
           input: toolCall.input as Record<string, unknown>,
           result,
@@ -4213,6 +4231,7 @@ export async function runAgentLoop(opts: {
         const result = planModeBlockedMessage(toolCall.name);
         send({
           type: "tool_done",
+          id: toolCall.id,
           tool: toolCall.name,
           input: toolCall.input as Record<string, unknown>,
           result,
@@ -4474,6 +4493,7 @@ export async function runAgentLoop(opts: {
 
       send({
         type: "tool_done",
+        id: toolCall.id,
         tool: toolCall.name,
         input: toolCall.input as Record<string, unknown>,
         result,