@agent-native/core 0.137.2 → 0.137.4

package/corpus/templates/macros/.agents/skills/frontend-design/SKILL.md

@@ -77,7 +77,10 @@ passed this requirement yet.
 
 ## Aesthetic Guidelines
 
-- **Typography**: Use the product's existing type system first. For net-new public pages, choose characterful but readable type and keep sizing appropriate to the surface.
+- **Typography**: Use the product's existing type system first. For SaaS and
+  operational apps, make a clear sans-serif hierarchy the default; reserve a
+  serif or editorial face for a deliberate content preview or brand moment,
+  not the whole application shell.
 - **Color and theme**: Use semantic tokens and CSS variables. Avoid one-note palettes and default purple/blue gradients unless the brand demands them.
 - **Motion**: Prefer purposeful transitions and small state changes. Use CSS transitions/keyframes unless the app already uses a motion library. Never `transition-all` — list the properties that actually change (e.g. `transition-[opacity,transform]`). Use the shared easing tokens defined in `packages/core/src/styles/agent-native.css` instead of hand-typing curves: `var(--ease-drawer)` (260ms, drawers/app chrome), `var(--ease-collapse)` (200ms, expand/collapse), `var(--ease-out-strong)` (snappy entrances) — in Tailwind, `ease-[var(--ease-collapse)]`. Enter/exit with ease-out, never `ease-in`. Overlays that zoom in must set the Radix origin var (e.g. `origin-[--radix-popover-content-transform-origin]`). Animate `transform`/`opacity`, not width/height/padding/box-shadow. Gate looping or large-movement animations with `motion-reduce:`. Command palettes and keyboard-triggered actions get no animation.
 - **Composition**: Match the workflow. Operational apps should be dense and scannable; marketing or portfolio pages can be more immersive.
@@ -115,6 +118,55 @@ passed this requirement yet.
 - Keep UI optimistic where possible: update cache and navigation immediately, then reconcile or roll back on mutation result.
 - Custom styles belong in Tailwind classes, component CSS, or the existing global CSS theme file; avoid inline styles.
 
+### Agent Surface And Page Boundaries
+
+- Treat the domain UI and the agent as two coordinated surfaces. The domain
+  page should make the repeatable job fast; the agent should handle judgment,
+  exploration, and actions that benefit from conversation.
+- Keep domain work on a named domain route such as `/automations`, `/block`, or
+  `/workflow`. Preserve the starter's full-page chat route (`/` or `/chat/*`)
+  when it exists; do not replace it with a domain form while leaving the shell
+  configured as if it were chat.
+- Use the persistent right `AgentSidebar` for contextual AI. A button that
+  sends work to `sendToAgentChat` must open or focus that sidebar and leave the
+  user on the current domain surface. Use full-page chat for chat-first work,
+  not as a hidden transport for a domain button.
+- Keep the left navigation domain-specific. A page called Automations,
+  Block time, or Create deck should not be nested under a generic Chat item;
+  Chat is its own destination and the right rail is the contextual assistant.
+- Never use sparkle, wand, magic, robot, or similar decorative AI icons. Use a
+  familiar message, assistant, or neutral action icon, and let the button copy
+  explain the intent. An icon-only control needs a tooltip and accessible name.
+- Give the persistent agent drawer a quiet but intentional boundary: a subtle
+  surface shift, divider, or both. The sidebar and domain page should not read
+  as one undifferentiated slab, and the treatment should remain calm when the
+  drawer opens or closes.
+- Treat an AI-labeled button as a contract. Buttons named Ask agent, Review
+  with agent, Refine, Generate with AI, or similar must call
+  `sendToAgentChat` with bounded context, `openSidebar: true`, and the intended
+  `submit` mode. A deterministic local action is useful, but label it local,
+  preview, or analyze rather than implying it invoked an agent.
+
+### Product Surface Review
+
+Before shipping a new app or a substantial redesign, review the first viewport
+as an operator would use it repeatedly:
+
+- Show one clear job title, the current state, and one primary next action.
+- Put optional inputs, provider choices, diagnostics, long explanations, and
+  secondary actions behind a disclosure, a later step, a menu, or the agent
+  sidebar. Do not put multiple unrelated forms side by side by default.
+- Prefer a short step flow or a focused page per domain job over one giant
+  dashboard. A route can have several states, but it should reveal one state
+  at a time and preserve the user's progress.
+- Remove generic hero copy, feature tours, repeated helper text, nested cards,
+  status-chip soup, and decorative AI treatment before adding more styling.
+- For review flows, stack the source/original first and the generated or safe
+  result second by default. Use side-by-side comparison only when the content
+  is short enough to scan without excessive horizontal reading.
+- Check the result at the target desktop width and a narrow width. If the first
+  viewport feels like documentation instead of a tool, subtract again.
+
 ## shadcn/ui Design Rules
 
 - Use built-in component variants first (`variant`, `size`) before overriding classes.

package/corpus/templates/mail/.agents/skills/agent-native-toolkit/SKILL.md

@@ -32,6 +32,39 @@ Move behavior into shared toolkit primitives when it is:
 Keep behavior app-local when the abstraction would hide important domain
 language or make a simple app-specific workflow harder to understand.
 
+## Agent Surface Contract
+
+The repeated app shell has two distinct navigation surfaces:
+
+- The left rail owns domain destinations, settings, and chat history when the
+  app has a full-page chat route. Do not label a domain workflow as `Chat` just
+  because the app was scaffolded from the chat template.
+- The right `AgentSidebar` owns contextual agent work. Domain buttons that call
+  `sendToAgentChat` should open it (`openSidebar: true`) so the user can see,
+  steer, and review the agent without losing the page they were using.
+- Keep `/` or `/chat/*` as the full-page chat surface when the starter provides
+  one. Put domain workflows on named routes and wire the shell's route checks,
+  navigation labels, and handoffs to those routes together.
+- Use familiar message or neutral action icons for agent affordances. Never use
+  sparkle, wand, magic, or robot icons as an AI label; the copy should carry the
+  meaning.
+- Give the right rail a quiet visual boundary with a subtle surface shift,
+  divider, or both. The domain page and AgentSidebar should not collapse into a
+  single undifferentiated background.
+- Any button labeled as agent work must use `sendToAgentChat` with bounded
+  context and `openSidebar: true`; local deterministic analysis should be
+  labeled as local, preview, or analyze. For original/generated review, stack
+  the source above the result by default and use side-by-side only for short,
+  highly scannable content.
+- Standalone apps with `AgentSidebar` must resolve one assistant-ui runtime
+  context. Match direct assistant-ui pins to the installed core/toolkit peer
+  graph, use Vite dedupe/aliases when linked dependencies can split contexts,
+  and verify an AI handoff produces no stale-index console error.
+
+Contextual agent UI is not a reason to expose every option at once. Start with
+the domain task's primary action, reveal review or configuration only when the
+current state needs it, and let the sidebar carry conversational depth.
+
 ## Discover Before Building
 
 Before creating an app-local version of repeated workspace or agent UI:

package/corpus/templates/mail/.agents/skills/frontend-design/SKILL.md

@@ -77,7 +77,10 @@ passed this requirement yet.
 
 ## Aesthetic Guidelines
 
-- **Typography**: Use the product's existing type system first. For net-new public pages, choose characterful but readable type and keep sizing appropriate to the surface.
+- **Typography**: Use the product's existing type system first. For SaaS and
+  operational apps, make a clear sans-serif hierarchy the default; reserve a
+  serif or editorial face for a deliberate content preview or brand moment,
+  not the whole application shell.
 - **Color and theme**: Use semantic tokens and CSS variables. Avoid one-note palettes and default purple/blue gradients unless the brand demands them.
 - **Motion**: Prefer purposeful transitions and small state changes. Use CSS transitions/keyframes unless the app already uses a motion library. Never `transition-all` — list the properties that actually change (e.g. `transition-[opacity,transform]`). Use the shared easing tokens defined in `packages/core/src/styles/agent-native.css` instead of hand-typing curves: `var(--ease-drawer)` (260ms, drawers/app chrome), `var(--ease-collapse)` (200ms, expand/collapse), `var(--ease-out-strong)` (snappy entrances) — in Tailwind, `ease-[var(--ease-collapse)]`. Enter/exit with ease-out, never `ease-in`. Overlays that zoom in must set the Radix origin var (e.g. `origin-[--radix-popover-content-transform-origin]`). Animate `transform`/`opacity`, not width/height/padding/box-shadow. Gate looping or large-movement animations with `motion-reduce:`. Command palettes and keyboard-triggered actions get no animation.
 - **Composition**: Match the workflow. Operational apps should be dense and scannable; marketing or portfolio pages can be more immersive.
@@ -115,6 +118,55 @@ passed this requirement yet.
 - Keep UI optimistic where possible: update cache and navigation immediately, then reconcile or roll back on mutation result.
 - Custom styles belong in Tailwind classes, component CSS, or the existing global CSS theme file; avoid inline styles.
 
+### Agent Surface And Page Boundaries
+
+- Treat the domain UI and the agent as two coordinated surfaces. The domain
+  page should make the repeatable job fast; the agent should handle judgment,
+  exploration, and actions that benefit from conversation.
+- Keep domain work on a named domain route such as `/automations`, `/block`, or
+  `/workflow`. Preserve the starter's full-page chat route (`/` or `/chat/*`)
+  when it exists; do not replace it with a domain form while leaving the shell
+  configured as if it were chat.
+- Use the persistent right `AgentSidebar` for contextual AI. A button that
+  sends work to `sendToAgentChat` must open or focus that sidebar and leave the
+  user on the current domain surface. Use full-page chat for chat-first work,
+  not as a hidden transport for a domain button.
+- Keep the left navigation domain-specific. A page called Automations,
+  Block time, or Create deck should not be nested under a generic Chat item;
+  Chat is its own destination and the right rail is the contextual assistant.
+- Never use sparkle, wand, magic, robot, or similar decorative AI icons. Use a
+  familiar message, assistant, or neutral action icon, and let the button copy
+  explain the intent. An icon-only control needs a tooltip and accessible name.
+- Give the persistent agent drawer a quiet but intentional boundary: a subtle
+  surface shift, divider, or both. The sidebar and domain page should not read
+  as one undifferentiated slab, and the treatment should remain calm when the
+  drawer opens or closes.
+- Treat an AI-labeled button as a contract. Buttons named Ask agent, Review
+  with agent, Refine, Generate with AI, or similar must call
+  `sendToAgentChat` with bounded context, `openSidebar: true`, and the intended
+  `submit` mode. A deterministic local action is useful, but label it local,
+  preview, or analyze rather than implying it invoked an agent.
+
+### Product Surface Review
+
+Before shipping a new app or a substantial redesign, review the first viewport
+as an operator would use it repeatedly:
+
+- Show one clear job title, the current state, and one primary next action.
+- Put optional inputs, provider choices, diagnostics, long explanations, and
+  secondary actions behind a disclosure, a later step, a menu, or the agent
+  sidebar. Do not put multiple unrelated forms side by side by default.
+- Prefer a short step flow or a focused page per domain job over one giant
+  dashboard. A route can have several states, but it should reveal one state
+  at a time and preserve the user's progress.
+- Remove generic hero copy, feature tours, repeated helper text, nested cards,
+  status-chip soup, and decorative AI treatment before adding more styling.
+- For review flows, stack the source/original first and the generated or safe
+  result second by default. Use side-by-side comparison only when the content
+  is short enough to scan without excessive horizontal reading.
+- Check the result at the target desktop width and a narrow width. If the first
+  viewport feels like documentation instead of a tool, subtract again.
+
 ## shadcn/ui Design Rules
 
 - Use built-in component variants first (`variant`, `size`) before overriding classes.

package/corpus/templates/mail/app/components/layout/AppLayout.tsx

@@ -1800,7 +1800,6 @@ function AppLayoutInner({ children }: AppLayoutProps) {
                 feedback={feedbackButton}
                 search={searchButton}
                 collapse={collapseButton}
-                className={showCollapsedSidebar ? undefined : "px-0 py-0"}
               />
             </div>
           </>
@@ -2193,7 +2192,6 @@ function StandardLayout({ children }: AppLayoutProps) {
               feedback={feedbackButton}
               search={searchButton}
               collapse={collapseButton}
-              className="px-0 py-0"
             />
           </div>
         </div>

package/corpus/templates/mail/app/pages/SettingsPage.tsx

@@ -1476,7 +1476,7 @@ export function SettingsPage() {
   const [searchParams, setSearchParams] = useSearchParams();
   const navState = useNavigationState();
   const agentSettingsTabs = useAgentSettingsTabs();
-  const [activeSection, setActiveSection] = useState<string>("general");
+  const [activeSection, setActiveSection] = useState<string>("integrations");
 
   const mailTabs = useMemo<SettingsTabItem[]>(
     () => [

package/corpus/templates/plan/.agents/skills/agent-native-toolkit/SKILL.md

@@ -32,6 +32,39 @@ Move behavior into shared toolkit primitives when it is:
 Keep behavior app-local when the abstraction would hide important domain
 language or make a simple app-specific workflow harder to understand.
 
+## Agent Surface Contract
+
+The repeated app shell has two distinct navigation surfaces:
+
+- The left rail owns domain destinations, settings, and chat history when the
+  app has a full-page chat route. Do not label a domain workflow as `Chat` just
+  because the app was scaffolded from the chat template.
+- The right `AgentSidebar` owns contextual agent work. Domain buttons that call
+  `sendToAgentChat` should open it (`openSidebar: true`) so the user can see,
+  steer, and review the agent without losing the page they were using.
+- Keep `/` or `/chat/*` as the full-page chat surface when the starter provides
+  one. Put domain workflows on named routes and wire the shell's route checks,
+  navigation labels, and handoffs to those routes together.
+- Use familiar message or neutral action icons for agent affordances. Never use
+  sparkle, wand, magic, or robot icons as an AI label; the copy should carry the
+  meaning.
+- Give the right rail a quiet visual boundary with a subtle surface shift,
+  divider, or both. The domain page and AgentSidebar should not collapse into a
+  single undifferentiated background.
+- Any button labeled as agent work must use `sendToAgentChat` with bounded
+  context and `openSidebar: true`; local deterministic analysis should be
+  labeled as local, preview, or analyze. For original/generated review, stack
+  the source above the result by default and use side-by-side only for short,
+  highly scannable content.
+- Standalone apps with `AgentSidebar` must resolve one assistant-ui runtime
+  context. Match direct assistant-ui pins to the installed core/toolkit peer
+  graph, use Vite dedupe/aliases when linked dependencies can split contexts,
+  and verify an AI handoff produces no stale-index console error.
+
+Contextual agent UI is not a reason to expose every option at once. Start with
+the domain task's primary action, reveal review or configuration only when the
+current state needs it, and let the sidebar carry conversational depth.
+
 ## Discover Before Building
 
 Before creating an app-local version of repeated workspace or agent UI:

package/corpus/templates/plan/.agents/skills/frontend-design/SKILL.md

@@ -77,7 +77,10 @@ passed this requirement yet.
 
 ## Aesthetic Guidelines
 
-- **Typography**: Use the product's existing type system first. For net-new public pages, choose characterful but readable type and keep sizing appropriate to the surface.
+- **Typography**: Use the product's existing type system first. For SaaS and
+  operational apps, make a clear sans-serif hierarchy the default; reserve a
+  serif or editorial face for a deliberate content preview or brand moment,
+  not the whole application shell.
 - **Color and theme**: Use semantic tokens and CSS variables. Avoid one-note palettes and default purple/blue gradients unless the brand demands them.
 - **Motion**: Prefer purposeful transitions and small state changes. Use CSS transitions/keyframes unless the app already uses a motion library. Never `transition-all` — list the properties that actually change (e.g. `transition-[opacity,transform]`). Use the shared easing tokens defined in `packages/core/src/styles/agent-native.css` instead of hand-typing curves: `var(--ease-drawer)` (260ms, drawers/app chrome), `var(--ease-collapse)` (200ms, expand/collapse), `var(--ease-out-strong)` (snappy entrances) — in Tailwind, `ease-[var(--ease-collapse)]`. Enter/exit with ease-out, never `ease-in`. Overlays that zoom in must set the Radix origin var (e.g. `origin-[--radix-popover-content-transform-origin]`). Animate `transform`/`opacity`, not width/height/padding/box-shadow. Gate looping or large-movement animations with `motion-reduce:`. Command palettes and keyboard-triggered actions get no animation.
 - **Composition**: Match the workflow. Operational apps should be dense and scannable; marketing or portfolio pages can be more immersive.
@@ -115,6 +118,55 @@ passed this requirement yet.
 - Keep UI optimistic where possible: update cache and navigation immediately, then reconcile or roll back on mutation result.
 - Custom styles belong in Tailwind classes, component CSS, or the existing global CSS theme file; avoid inline styles.
 
+### Agent Surface And Page Boundaries
+
+- Treat the domain UI and the agent as two coordinated surfaces. The domain
+  page should make the repeatable job fast; the agent should handle judgment,
+  exploration, and actions that benefit from conversation.
+- Keep domain work on a named domain route such as `/automations`, `/block`, or
+  `/workflow`. Preserve the starter's full-page chat route (`/` or `/chat/*`)
+  when it exists; do not replace it with a domain form while leaving the shell
+  configured as if it were chat.
+- Use the persistent right `AgentSidebar` for contextual AI. A button that
+  sends work to `sendToAgentChat` must open or focus that sidebar and leave the
+  user on the current domain surface. Use full-page chat for chat-first work,
+  not as a hidden transport for a domain button.
+- Keep the left navigation domain-specific. A page called Automations,
+  Block time, or Create deck should not be nested under a generic Chat item;
+  Chat is its own destination and the right rail is the contextual assistant.
+- Never use sparkle, wand, magic, robot, or similar decorative AI icons. Use a
+  familiar message, assistant, or neutral action icon, and let the button copy
+  explain the intent. An icon-only control needs a tooltip and accessible name.
+- Give the persistent agent drawer a quiet but intentional boundary: a subtle
+  surface shift, divider, or both. The sidebar and domain page should not read
+  as one undifferentiated slab, and the treatment should remain calm when the
+  drawer opens or closes.
+- Treat an AI-labeled button as a contract. Buttons named Ask agent, Review
+  with agent, Refine, Generate with AI, or similar must call
+  `sendToAgentChat` with bounded context, `openSidebar: true`, and the intended
+  `submit` mode. A deterministic local action is useful, but label it local,
+  preview, or analyze rather than implying it invoked an agent.
+
+### Product Surface Review
+
+Before shipping a new app or a substantial redesign, review the first viewport
+as an operator would use it repeatedly:
+
+- Show one clear job title, the current state, and one primary next action.
+- Put optional inputs, provider choices, diagnostics, long explanations, and
+  secondary actions behind a disclosure, a later step, a menu, or the agent
+  sidebar. Do not put multiple unrelated forms side by side by default.
+- Prefer a short step flow or a focused page per domain job over one giant
+  dashboard. A route can have several states, but it should reveal one state
+  at a time and preserve the user's progress.
+- Remove generic hero copy, feature tours, repeated helper text, nested cards,
+  status-chip soup, and decorative AI treatment before adding more styling.
+- For review flows, stack the source/original first and the generated or safe
+  result second by default. Use side-by-side comparison only when the content
+  is short enough to scan without excessive horizontal reading.
+- Check the result at the target desktop width and a narrow width. If the first
+  viewport feels like documentation instead of a tool, subtract again.
+
 ## shadcn/ui Design Rules
 
 - Use built-in component variants first (`variant`, `size`) before overriding classes.

package/corpus/templates/slides/.agents/skills/agent-native-toolkit/SKILL.md

@@ -32,6 +32,39 @@ Move behavior into shared toolkit primitives when it is:
 Keep behavior app-local when the abstraction would hide important domain
 language or make a simple app-specific workflow harder to understand.
 
+## Agent Surface Contract
+
+The repeated app shell has two distinct navigation surfaces:
+
+- The left rail owns domain destinations, settings, and chat history when the
+  app has a full-page chat route. Do not label a domain workflow as `Chat` just
+  because the app was scaffolded from the chat template.
+- The right `AgentSidebar` owns contextual agent work. Domain buttons that call
+  `sendToAgentChat` should open it (`openSidebar: true`) so the user can see,
+  steer, and review the agent without losing the page they were using.
+- Keep `/` or `/chat/*` as the full-page chat surface when the starter provides
+  one. Put domain workflows on named routes and wire the shell's route checks,
+  navigation labels, and handoffs to those routes together.
+- Use familiar message or neutral action icons for agent affordances. Never use
+  sparkle, wand, magic, or robot icons as an AI label; the copy should carry the
+  meaning.
+- Give the right rail a quiet visual boundary with a subtle surface shift,
+  divider, or both. The domain page and AgentSidebar should not collapse into a
+  single undifferentiated background.
+- Any button labeled as agent work must use `sendToAgentChat` with bounded
+  context and `openSidebar: true`; local deterministic analysis should be
+  labeled as local, preview, or analyze. For original/generated review, stack
+  the source above the result by default and use side-by-side only for short,
+  highly scannable content.
+- Standalone apps with `AgentSidebar` must resolve one assistant-ui runtime
+  context. Match direct assistant-ui pins to the installed core/toolkit peer
+  graph, use Vite dedupe/aliases when linked dependencies can split contexts,
+  and verify an AI handoff produces no stale-index console error.
+
+Contextual agent UI is not a reason to expose every option at once. Start with
+the domain task's primary action, reveal review or configuration only when the
+current state needs it, and let the sidebar carry conversational depth.
+
 ## Discover Before Building
 
 Before creating an app-local version of repeated workspace or agent UI:

package/corpus/templates/slides/.agents/skills/frontend-design/SKILL.md

@@ -77,7 +77,10 @@ passed this requirement yet.
 
 ## Aesthetic Guidelines
 
-- **Typography**: Use the product's existing type system first. For net-new public pages, choose characterful but readable type and keep sizing appropriate to the surface.
+- **Typography**: Use the product's existing type system first. For SaaS and
+  operational apps, make a clear sans-serif hierarchy the default; reserve a
+  serif or editorial face for a deliberate content preview or brand moment,
+  not the whole application shell.
 - **Color and theme**: Use semantic tokens and CSS variables. Avoid one-note palettes and default purple/blue gradients unless the brand demands them.
 - **Motion**: Prefer purposeful transitions and small state changes. Use CSS transitions/keyframes unless the app already uses a motion library. Never `transition-all` — list the properties that actually change (e.g. `transition-[opacity,transform]`). Use the shared easing tokens defined in `packages/core/src/styles/agent-native.css` instead of hand-typing curves: `var(--ease-drawer)` (260ms, drawers/app chrome), `var(--ease-collapse)` (200ms, expand/collapse), `var(--ease-out-strong)` (snappy entrances) — in Tailwind, `ease-[var(--ease-collapse)]`. Enter/exit with ease-out, never `ease-in`. Overlays that zoom in must set the Radix origin var (e.g. `origin-[--radix-popover-content-transform-origin]`). Animate `transform`/`opacity`, not width/height/padding/box-shadow. Gate looping or large-movement animations with `motion-reduce:`. Command palettes and keyboard-triggered actions get no animation.
 - **Composition**: Match the workflow. Operational apps should be dense and scannable; marketing or portfolio pages can be more immersive.
@@ -115,6 +118,55 @@ passed this requirement yet.
 - Keep UI optimistic where possible: update cache and navigation immediately, then reconcile or roll back on mutation result.
 - Custom styles belong in Tailwind classes, component CSS, or the existing global CSS theme file; avoid inline styles.
 
+### Agent Surface And Page Boundaries
+
+- Treat the domain UI and the agent as two coordinated surfaces. The domain
+  page should make the repeatable job fast; the agent should handle judgment,
+  exploration, and actions that benefit from conversation.
+- Keep domain work on a named domain route such as `/automations`, `/block`, or
+  `/workflow`. Preserve the starter's full-page chat route (`/` or `/chat/*`)
+  when it exists; do not replace it with a domain form while leaving the shell
+  configured as if it were chat.
+- Use the persistent right `AgentSidebar` for contextual AI. A button that
+  sends work to `sendToAgentChat` must open or focus that sidebar and leave the
+  user on the current domain surface. Use full-page chat for chat-first work,
+  not as a hidden transport for a domain button.
+- Keep the left navigation domain-specific. A page called Automations,
+  Block time, or Create deck should not be nested under a generic Chat item;
+  Chat is its own destination and the right rail is the contextual assistant.
+- Never use sparkle, wand, magic, robot, or similar decorative AI icons. Use a
+  familiar message, assistant, or neutral action icon, and let the button copy
+  explain the intent. An icon-only control needs a tooltip and accessible name.
+- Give the persistent agent drawer a quiet but intentional boundary: a subtle
+  surface shift, divider, or both. The sidebar and domain page should not read
+  as one undifferentiated slab, and the treatment should remain calm when the
+  drawer opens or closes.
+- Treat an AI-labeled button as a contract. Buttons named Ask agent, Review
+  with agent, Refine, Generate with AI, or similar must call
+  `sendToAgentChat` with bounded context, `openSidebar: true`, and the intended
+  `submit` mode. A deterministic local action is useful, but label it local,
+  preview, or analyze rather than implying it invoked an agent.
+
+### Product Surface Review
+
+Before shipping a new app or a substantial redesign, review the first viewport
+as an operator would use it repeatedly:
+
+- Show one clear job title, the current state, and one primary next action.
+- Put optional inputs, provider choices, diagnostics, long explanations, and
+  secondary actions behind a disclosure, a later step, a menu, or the agent
+  sidebar. Do not put multiple unrelated forms side by side by default.
+- Prefer a short step flow or a focused page per domain job over one giant
+  dashboard. A route can have several states, but it should reveal one state
+  at a time and preserve the user's progress.
+- Remove generic hero copy, feature tours, repeated helper text, nested cards,
+  status-chip soup, and decorative AI treatment before adding more styling.
+- For review flows, stack the source/original first and the generated or safe
+  result second by default. Use side-by-side comparison only when the content
+  is short enough to scan without excessive horizontal reading.
+- Check the result at the target desktop width and a narrow width. If the first
+  viewport feels like documentation instead of a tool, subtract again.
+
 ## shadcn/ui Design Rules
 
 - Use built-in component variants first (`variant`, `size`) before overriding classes.

package/corpus/templates/slides/actions/generate-image-api.ts

@@ -8,7 +8,10 @@ import {
   extractAssetUrl,
   imagePreviewMarkdown,
 } from "../server/lib/assets-image-delegation.js";
-import { DEFAULT_STYLE_REFERENCE_URLS } from "../shared/api.js";
+import {
+  DEFAULT_STYLE_REFERENCE_URLS,
+  normalizeReferenceUrls,
+} from "../shared/api.js";
 
 interface ReferenceImage {
   data: string; // base64
@@ -47,6 +50,10 @@ export default defineAction({
       .string()
       .optional()
       .describe("HTML of the target slide, used as generation context"),
+    referenceImageUrls: z
+      .array(z.string())
+      .optional()
+      .describe("Style reference URLs to forward to Assets"),
   }),
   run: async (args) => {
     const prompt = args.prompt;
@@ -59,6 +66,7 @@ export default defineAction({
       deckId: args.deckId,
       slideId: args.slideId,
       slideContent: args.slideContent,
+      referenceImageUrls: normalizeReferenceUrls(args.referenceImageUrls),
     });
 
     if (delegation.status === "delegated") {
@@ -78,7 +86,7 @@ export default defineAction({
     if (delegation.status === "pending") {
       throw new Error(
         `Assets is still generating (task ${delegation.taskId}, state "${delegation.lastState}"). ` +
-          `It was not cancelled — check the Assets app for the result instead of generating again.`,
+          `It was not cancelled - check the Assets app for the result instead of generating again.`,
       );
     }
 
@@ -88,7 +96,7 @@ export default defineAction({
       );
     }
 
-    // Assets is unreachable — standalone-deploy fallback. The caller is told
+    // Assets is unreachable - standalone-deploy fallback. The caller is told
     // which path ran and why, so a brand-inconsistent image is never reported
     // as a library-grounded one.
     const { getProvider } =
@@ -97,7 +105,10 @@ export default defineAction({
 
     const refImages: ReferenceImage[] = [];
     const results = await Promise.all(
-      DEFAULT_STYLE_REFERENCE_URLS.map(urlToReferenceImage),
+      normalizeReferenceUrls([
+        ...DEFAULT_STYLE_REFERENCE_URLS,
+        ...normalizeReferenceUrls(args.referenceImageUrls),
+      ]).map(urlToReferenceImage),
     );
     for (const r of results) {
       if (r) refImages.push(r);

package/corpus/templates/slides/actions/import-google-slides-reference.ts

@@ -0,0 +1,83 @@
+import { defineAction } from "@agent-native/core";
+import { buildDeepLink } from "@agent-native/core/server";
+import { getRequestUserEmail } from "@agent-native/core/server/request-context";
+import { z } from "zod";
+
+import { getGoogleDocsAccessToken } from "../server/lib/google-docs-oauth.js";
+import { importPptxBufferToDeck } from "./import-pptx.js";
+
+function deckDeepLink(deckId: string): string {
+  return buildDeepLink({
+    app: "slides",
+    view: "editor",
+    params: { deckId },
+  });
+}
+
+async function exportGoogleSlidesAsPptx(fileId: string, accessToken: string) {
+  const response = await fetch(
+    `https://www.googleapis.com/drive/v3/files/${encodeURIComponent(fileId)}/export?mimeType=application/vnd.openxmlformats-officedocument.presentationml.presentation`,
+    {
+      headers: { Authorization: `Bearer ${accessToken}` },
+    },
+  );
+
+  if (!response.ok) {
+    throw new Error(
+      `Google Drive returned HTTP ${response.status} while exporting the presentation.`,
+    );
+  }
+
+  const bytes = await response.arrayBuffer();
+  if (bytes.byteLength === 0) {
+    throw new Error("Google Drive returned an empty presentation export.");
+  }
+
+  return Buffer.from(bytes);
+}
+
+export default defineAction({
+  description:
+    "Import a Google Slides deck selected from Google Picker and save it as a reusable Slides reference deck. " +
+    "The selected presentation is exported from Drive as PPTX, parsed, and stored as a normal deck in the caller's Slides workspace.",
+  schema: z.object({
+    fileId: z.string().describe("Google Slides file ID from Google Picker"),
+    title: z
+      .string()
+      .optional()
+      .describe("Optional title for the imported reference deck"),
+  }),
+  run: async ({ fileId, title }) => {
+    const owner = getRequestUserEmail();
+    if (!owner) throw new Error("no authenticated user");
+
+    const connection = await getGoogleDocsAccessToken(owner);
+    if (!connection) {
+      throw new Error(
+        "Connect Google Docs before importing a Google Slides deck.",
+      );
+    }
+
+    const fileBuffer = await exportGoogleSlidesAsPptx(
+      fileId,
+      connection.accessToken,
+    );
+    return importPptxBufferToDeck({
+      fileBuffer,
+      title,
+      source: "import-google-slides-reference",
+    });
+  },
+  link: ({ result }) => {
+    const id =
+      result && typeof result === "object"
+        ? (result as { id?: string }).id
+        : undefined;
+    if (!id) return null;
+    return {
+      url: deckDeepLink(id),
+      label: "Open deck in Slides",
+      view: "editor",
+    };
+  },
+});