@lowdefy/docs-content 0.0.0-experimental-20260928120643 → 6.1.0

package/content/concepts/events-and-actions.md

@@ -256,11 +256,9 @@ events:
           searchOpen: true
 ```
 
-## Page lifecycle events
+## Page initialization events
 
-The first blocks on a page, usually a [`container`](/container) type block, can define page lifecycle events. The page initialization events `onInit` and `onInitAsync` are triggered when the page is first loaded, and the browser lifecycle events `onVisible`, `onHidden`, `onOnline`, `onOffline` and `onResize` are triggered while the page stays open. All blocks have `onMount` and `onMountAsync` events, that can be used to initialize the page or blocks.
-
-### Page initialization events
+The first blocks on a page, usually a [`container`](/container) type block, can define `onInit` and `onInitAsync` events. All blocks have `onMount` and `onMountAsync` events, that can be used to initialize the page or blocks.
 
 The `onInit` event is triggered the first time a page is loaded. This event blocks page render, in other words, the page __will__ remain in a loading state, rendering only the progress bar, until all the actions have completed execution. It can be used to set up [`state`](/page-and-app-state). Actions that take a long time to execute, like `Request`, should be used sparingly here for a better user experience.
 
@@ -270,62 +268,6 @@ The `onMount` event is triggered every time a block is rendered on a page. This
 
 The `onMountAsync` event is triggered every time a block is mounted, but does not render the block in loading.
 
-### Browser lifecycle events
-
-The first blocks on a page can also define events that react to the browser: `onVisible`, `onHidden`, `onOnline`, `onOffline` and `onResize`. These are useful to refresh data that went stale while the user was away, instead of polling the server on a timer.
-
-These events are only attached while the page is mounted, and are removed again when the user navigates away, so they never trigger for a page the user has left. They never trigger on the initial page load either — only a change after the page has loaded triggers them. Their action chains run like `onMountAsync`: they do not keep the page or block in a loading state.
-
-The `onVisible` event is triggered when the browser tab becomes visible again, or when the window regains focus. Rapid pairs of these signals — a tab switch usually produces both — are coalesced over 300ms, so the event is only triggered once. The [`_event`](/_event) object contains `visible: true` and `reason`, which is `visibility` when the browser reported the tab becoming visible, or `focus` when the window regained focus.
-
-```yaml
-events:
-  onVisible:
-    - id: refresh_tasks
-      type: Request
-      params: get_tasks
-```
-
-The `onHidden` event is triggered when the browser tab is hidden, or when the window loses focus, coalesced in the same way. The `_event` object contains `visible: false` and the same `reason` field.
-
-```yaml
-events:
-  onHidden:
-    - id: stop_refreshing
-      type: SetState
-      params:
-        live: false
-```
-
-The `onOnline` and `onOffline` events are triggered when the browser reports that the network connection was regained or lost. The `_event` object contains `online: true` or `online: false`.
-
-```yaml
-events:
-  onOnline:
-    - id: back_online
-      type: Message
-      params:
-        content: You are back online.
-  onOffline:
-    - id: went_offline
-      type: Message
-      params:
-        status: warning
-        content: You are offline, changes will not be saved.
-```
-
-The `onResize` event is triggered when the window is resized, debounced by 200ms so that the actions only run once the user stops resizing. The `_event` object contains the `width` and `height` of the window.
-
-```yaml
-events:
-  onResize:
-    - id: set_width
-      type: SetState
-      params:
-        window_width:
-          _event: width
-```
-
 ## Action types
 
 The following actions can be used:
@@ -358,7 +300,3 @@ See additional action type available under the Actions tab in the menu.
 - The `onInitAsync` event is triggered the first time a page is mounted and does not keep the page in loading.
 - The `onMount` events is triggered the every time a block is mounted and keeps the block in loading until all actions have finished.
 - The `onMountAsync` event is triggered the every time a block is mounted, after `onMount` has completed, and does not keep the block in loading.
-- The `onVisible` and `onHidden` events are triggered when the browser tab is shown or hidden, or when the window gains or loses focus, coalesced over 300ms.
-- The `onOnline` and `onOffline` events are triggered when the browser reports that the network connection was regained or lost.
-- The `onResize` event is triggered when the window is resized, debounced by 200ms.
-- Browser lifecycle events are never triggered on the initial page load, and are removed when the user navigates away from the page.

package/content/concepts/i18n.md

@@ -162,17 +162,12 @@ config:
 | `client.requestError` | `Request error` | Fallback when an API response has no message |
 | `client.backAriaLabel` | `back` | Aria label on browser-back links |
 | `client.shortcutThen` | `then` | Sequence shortcut badge separator |
-| `client.copy` | `Copy` | Label of a `data-copy` button in HTML |
-| `client.copyValue` | `Copy: {value}` | Label of a `data-copy` button whose value differs from its text |
-| `client.copied` | `Copied` | Announced after a `data-copy` copy |
-| `client.copyFailed` | `Copy failed` | Announced when a `data-copy` copy fails |
-| `client.confirm` | `Are you sure?` | Message of a `data-confirm` with no value |
 | `errorPage.name` | `Error` | Full-page error fallback heading |
 | `errorPage.message` | `An error has occurred.` | Full-page error fallback body |
 | `errorPage.homeLink` | `Return to home page` | Full-page error fallback link |
 | `server.genericError` | `Something went wrong.` | Message of any server error that is not an author error or an auth refusal |
 
-`server.genericError` is the message users see for any server error that is not an author error (`:throw`, `:reject`, a failed `ValidateSchema` step, a plugin's `UserError`) or an authentication or authorization refusal. The server translates it for the request's locale, the same way in development and production, and it is the message that `_actions`, `_request_details` and [`_error`](/_error) read in page config. To show something else for one action, set `messages.error` on the action to override its error toast. To show the real message, read it with [`_error`](/_error) inside the endpoint's `:catch` and send it on, for example `:throw: { _error: message }`.
+`server.genericError` is the message users see for any server error that is not an author error (`:throw`, `:reject`, a failed `ValidateSchema` step, a plugin's `UserError`) or an authentication refusal. The server translates it for the request's locale, the same way in development and production, and it is the message that `_actions`, `_request_details` and [`_error`](/_error) read in page config. To show something else for one action, set `messages.error` on the action to override its error toast. To show the real message, read it with [`_error`](/_error) inside the endpoint's `:catch` and send it on, for example `:throw: { _error: message }`.
 
 ## Localizing block defaults
 
@@ -223,7 +218,7 @@ config:
         agent.antdx.actions.feedbackDislike: 'Gefällt mir nicht'
 ```
 
-**Agent runtime errors.** HTTP error responses from the agent endpoint (a 405 for a wrong method, a 400 for a malformed request, and so on: `Only POST requests are supported.`, `Invalid agent path`, `Agent "X" does not exist.`, etc.) translate via `agent.runtime.*` keys against the request's `Accept-Language`. Server-side validation errors and stack traces stay in English (developer diagnostics).
+**Agent runtime errors.** HTTP 400/500 errors from the agent endpoint (`Only POST requests are supported.`, `Invalid agent path`, `Agent "X" does not exist.`, etc.) translate via `agent.runtime.*` keys against the request's `Accept-Language`. Server-side validation errors and stack traces stay in English (developer diagnostics).
 
 **Agent config translation.** `agent.properties.instructions` passes through the operator parser at request time, so `_t:` works:
 

package/content/concepts/keyboard-shortcuts.md

@@ -139,7 +139,7 @@ When a shortcut matches, the browser's default action for that key combination i
   type: Button
   properties:
     title: Search
-    icon: search
+    icon: AiOutlineSearch
   events:
     onClick:
       shortcut: mod+K
@@ -157,7 +157,7 @@ When a shortcut matches, the browser's default action for that key combination i
   type: Button
   properties:
     title: Save
-    icon: save
+    icon: AiOutlineSave
   events:
     onClick:
       shortcut: mod+S

package/content/concepts/layout-overview.md

@@ -103,7 +103,7 @@ Blocks can also be laid out using [CSS flexbox](https://developer.mozilla.org/en
 The `layout` object on blocks can be used to control how a block is placed in the layout. The `layout` properties that can be defined are:
 
 - `selfAlign`: _Enum_ - Align block vertically in the area. Options are `top`, `middle`, and `bottom`. Default `top.`
-- `align`: _Enum_ - Align content area children vertically. Options are `top`, `middle`, and `bottom`. Default `top.` This value flows to the content area as a default. In Lowdefy v4 `layout.align` aligned the block itself within its parent — that is now `layout.selfAlign`. Setting `layout.align` on a block aligns the block's own content, never the block itself.
+- `align`: _Enum_ - Align content area children vertically. Options are `top`, `middle`, and `bottom`. Default `top.` This value flows to the content area as a default.
 - `flex`: _String_ - Set the [`flex`](https://developer.mozilla.org/en-US/docs/Web/CSS/flex) CSS property. This overwrites the `grow`, `shrink`, and `size` properties.
 - `gap`: _Number_ | _Array_ - Set the gap (space) between blocks in the content area. If an array, the first element is the horizontal gap, and the second is the vertical gap. This value flows to the content area as a default.
 - `grow`: _Number_ - Set the [`flex-grow`](https://developer.mozilla.org/en-US/docs/Web/CSS/flex-grow) CSS property. Default 0.

package/content/concepts/logger.md

@@ -1,86 +1,5 @@
 # Logger
 
-Lowdefy provides built-in error logging and monitoring capabilities. When properly configured, errors from both client and server are captured and sent to external logging services.
-
-## Sentry Integration
-
-[Sentry](https://sentry.io) is a popular error tracking and performance monitoring platform. Lowdefy has built-in Sentry integration that captures errors with rich context including page IDs, block IDs, and config file locations.
-
-### Quick Start
-
-To enable Sentry, set the `SENTRY_DSN` environment variable:
-
-```
-SENTRY_DSN=https://your-dsn@sentry.io/project
-```
-
-With this environment variable set, Sentry will capture errors from both client and server with sensible defaults.
-
-### Client-Side Configuration
-
-For client-side error capture, you also need to expose the DSN to the browser:
-
-```
-```
-
-### Configuration Options
-
-You can customize Sentry behavior using the `logger` configuration in your `lowdefy.yaml`:
-
-```yaml
-lowdefy: 6.0.0
-
-logger:
-  sentry:
-    # Enable/disable client-side logging (default: true)
-    client: true
-    # Enable/disable server-side logging (default: true)
-    server: true
-    # Sample rate for performance traces (default: 0.1)
-    tracesSampleRate: 0.1
-    # Sample rate for session replay (default: 0)
-    replaysSessionSampleRate: 0
-    # Sample rate for replay on errors (default: 0.1)
-    replaysOnErrorSampleRate: 0.1
-    # Enable user feedback widget (default: false)
-    feedback: false
-    # Override environment detection (default: auto from NODE_ENV)
-    environment: production
-    # User fields to include in error reports (default: ['id', '_id'])
-    userFields:
-      - id
-      - _id
-```
-
-### Default Configuration
-
-If you only set `SENTRY_DSN` without any `logger.sentry` configuration, these defaults are used:
-
-| Option | Default | Description |
-|--------|---------|-------------|
-| `client` | `true` | Client-side error capture enabled |
-| `server` | `true` | Server-side error capture enabled |
-| `tracesSampleRate` | `0.1` | 10% of transactions traced |
-| `replaysSessionSampleRate` | `0` | Session replay disabled by default |
-| `replaysOnErrorSampleRate` | `0.1` | 10% of error sessions replayed |
-| `feedback` | `false` | User feedback widget disabled |
-| `userFields` | `['id', '_id']` | Only user ID fields logged |
-
-### User Context
-
-For authenticated users, Sentry automatically captures user context based on the `userFields` configuration. By default, only `id` and `_id` fields from the user session are included to avoid logging PII (personally identifiable information).
-
-To include additional fields:
-
-```yaml
-logger:
-  sentry:
-    userFields:
-      - id
-      - _id
-      - organization_id
-```
-
 ### Source Maps
 
 To enable readable stack traces in Sentry, set the `SENTRY_AUTH_TOKEN` environment variable during build:

package/content/concepts/lowdefy-schema.md

@@ -42,20 +42,17 @@ Find the more detailed block schema [here](/blocks).
 The config object has the following properties:
 
 - `environments: object`: The deployment environments of the app, keyed by name (for example `prod`, `staging`, `local`). Each build is for one of them, named by the `LOWDEFY_ENVIRONMENT` environment variable, and the current environment supplies the defaults for everything environment-specific. See [Deployment environments](/deployment-environments). Each environment has:
-  - `url: string`: The deployment origin, for example `https://staging.example.com`. Default for the auth canonical URL (`BETTER_AUTH_URL`) and the MCP resource URIs, notification links and logos (`RenderNotification` `serverUrl`), and where crons are forwarded to.
+  - `url: string`: The deployment origin, for example `https://staging.example.com`. Default for the auth base URL (`AUTH_URL`), notification links and logos (`RenderNotification` `serverUrl`), and where crons are forwarded to.
   - `cron.secret: string`: The **name** of the Lowdefy secret holding this environment's `CRON_SECRET` — a plain string, not a `_secret` operator (environment settings are resolved at build time, where `_secret` is not evaluated). The value is read at runtime from `LOWDEFY_SECRET_<name>` on the deployment that forwards. An environment with a `cron.secret` has its schedules forwarded to it from the environment Vercel fires crons on.
   - `cron.enabled: boolean`: Set `false` to register no cron jobs for this environment. Defaults to `true`.
   - `email.filter: object`: The delivery filter (`replaceAddress`, `allowlist`, `regex`) applied to every `SMTPMailSend` and `SendGridMailSend` request in this environment, unless the connection sets its own `filter` (`filter: false` on a connection turns filtering off).
   - `email.enabled: boolean`: Set `false` to send no email from `SMTPMailSend` and `SendGridMailSend` requests in this environment (auth emails still send). Defaults to `true`.
   - `sentry.enabled: boolean`: Set `false` to turn Sentry off, server and client, in this environment. Defaults to `true`.
   - `guards: object`: Pin secrets and environment variables to a pattern kept in config. `guards.secrets` maps a Lowdefy secret name (`LOWDEFY_SECRET_<name>`) and `guards.env` an environment variable name to a regular expression; in this environment the build fails unless each value matches (a missing value fails too), so changing a guarded value also takes a config change. See [Guards](/deployment-environments).
-  - `posthog.enabled: boolean`: Set `false` to turn PostHog off in this environment: `PostHogInit` behaves as `enabled: false` (no key needed) and every PostHog action is a no-op. Defaults to `true`.
 - `environment: string`: The name of the environment this build is for, used when the `LOWDEFY_ENVIRONMENT` environment variable is not set; the variable always wins. When `environments` is declared it must be one of its names. Server log lines carry it as `environment`, `_app: environment` reads it, and Sentry reports under it unless `logger.sentry.environment` is set.
 - `basePath: string`: Set the base path to serve the Lowdefy application from. This will route all pages under `https://example.com/<base-path>/<page-id>` instead of the default `https://example.com/<page-id>`. The basePath value must start with "/".
 - `homePageId: string`: The id of the page that will load when a visitor navigates to the home route, in other words, when the visitor navigates to `yourdomain.com` or `yourdomain.com/`.
-- `dependencyTracking: boolean`: Defaults to `true`. The browser re-evaluates only the blocks whose inputs changed (the state paths, requests and globals their operators read) instead of every block on the page after each change. Set to `false` to evaluate every block on every update, as earlier versions did.
 - `requestTimeout: number`: Maximum time in milliseconds a request may run before the server returns a timeout. This protects against requests that hang on an upstream call (database, SMTP, an external API) running all the way to the host's function limit — important on serverless platforms that bill by execution duration. Defaults to `30000` (30 seconds). Set to `0` to disable. Agent streaming routes are exempt, since they are long-lived by design.
-- `trustedProxies: string[]`: IP addresses and CIDR ranges (for example `10.0.0.0/8`) of the reverse proxies and load balancers in front of a self-hosted server. The server takes each request's client address from its connection, and reads `X-Forwarded-For` only when that connection comes from a trusted proxy, skipping trusted hops from the right. Auth rate limits, sessions and request logs use this address. Defaults to none, so a client cannot choose its own address. Not used on Vercel, where the platform supplies the client address. See [Behind a reverse proxy](/docker#behind-a-reverse-proxy).
 
 ## Auth
 

package/content/concepts/menus.md

@@ -24,7 +24,7 @@ The schema for a `MenuLink` is:
 - `style: object`: CSS styles for the item. Flat applies to the wrapper; use dot-prefixed slot keys (`.element`, `.icon`, `.label`) to target specific parts.
 -  `properties: object`: The properties of the menu link. These are:
     - `title: string`: The title to display for the link.
-    - `icon: string | object`: An icon name, such as the semantic name `home` or the [Lucide](https://lucide.dev/icons) name `House`, or properties of an [Icon](/Icon) block to use as the icon for the link.
+    - `icon: string | object`: The name of a [React Icon](https://react-icons.github.io/react-icons/search) or properties of an Icon block to use as the icon for the link.
     - `danger: boolean`: Apply danger style (switches the item onto antd's `dangerItem*` token set).
     - `disabled: boolean`: Disable the item — blocks clicks and applies a greyed style.
     - `tooltip: string`: Tooltip text shown when the menu is collapsed.
@@ -38,7 +38,7 @@ The schema for a `MenuGroup` is:
 - `style: object`: CSS styles for the group's title row. Use dot-prefixed slot keys to target specific parts.
 -  `properties: object`: The properties of the menu group. These are:
     - `title: string`: The title to display for the group.
-    - `icon: string | object`: An icon name, such as the semantic name `home` or the [Lucide](https://lucide.dev/icons) name `House`, or properties of an [Icon](/Icon) block to use as the icon for the group.
+    - `icon: string | object`: The name of a [React Icon](https://react-icons.github.io/react-icons/search) or properties of an Icon block to use as the icon for the group.
     - `disabled: boolean`: Disable the group — blocks expansion and applies a greyed style.
     - `tooltip: string`: Tooltip text shown when the menu is collapsed.
 - `links: object[]`: An array of `MenuLink` or `MenuGroup` objects that should be grouped together in the group.
@@ -62,11 +62,11 @@ menus:
         pageId: page1
         properties:
           title: Page 1
-          icon: file
+          icon: AiOutlineFile
       - id: top-group
         properties:
           title: Group
-          icon: Group
+          icon: AiOutlineGroup
         links:
           - id: page2
             type: MenuLink
@@ -93,13 +93,13 @@ menus:
         pageId: page1
         properties:
           title: Page 1
-          icon: file
+          icon: AiOutlineFile
       - id: page3
         type: MenuLink
         pageId: page3
         properties:
           title: Page 3
-          icon: SlidersHorizontal
+          icon: AiOutlineControl
 ```
 
 We can make use of [protected pages and APIs](/protected-pages-apis) and [roles](/roles) to filter the menu items according to the pages that the user has access to.
@@ -130,7 +130,7 @@ menus:
         pageId: page1
         properties:
           title: Page 1
-          icon: file
+          icon: AiOutlineFile
       - id: page2
         type: MenuLink
         pageId: page2 # pageId will be used as link title

package/content/concepts/module-authoring.md

@@ -416,14 +416,14 @@ links:
       _module.pageId: users-list
     properties:
       title: Users
-      icon: user
+      icon: AiOutlineUser
   - id: invite
     type: MenuLink
     pageId:
       _module.pageId: user-invite
     properties:
       title: Invite User
-      icon: UserPlus
+      icon: AiOutlineUserAdd
 ```
 
 The app wraps these in a MenuGroup:

package/content/connections/anthropic.md

@@ -11,8 +11,6 @@ The Anthropic connection connects to the [Anthropic API](https://docs.anthropic.
 #### Properties
 - `apiKey: string`: __Required__ - Anthropic API key. Use [`_secret`](/_secret) to reference this securely.
 - `baseURL: string`: Optional base URL for the Anthropic API. Use this to connect to a proxy or compatible endpoint.
-- `maxOutputTokens: number`: Default maximum number of tokens a model call generates, for the requests and agents on this connection that do not set their own.
-- `timeout: number`: Default milliseconds a model call may take, retries included, before it is cancelled, for the requests and agents on this connection that do not set their own. See _Limits and cancellation_ below.
 
 ###### Connection example:
 ```yaml
@@ -93,28 +91,11 @@ Request types:
 
 - GenerateText
 - GenerateObject
-- Decide
 
-All AI provider connections (`Anthropic`, `OpenAI`, `Google`, `AIGateway`) provide the same `GenerateText`, `GenerateObject` and `Decide` request types — the `connectionId` selects the provider. These make a single, one-shot model call and return the result. They can be used as page requests, or as steps in [API endpoint routines](/api) — useful for classify, extract, summarize, or decide steps where the surrounding logic stays in your routine. For a closed-vocabulary decision — pick one of these options, yes or no, rate on this scale — use `Decide`.
+All AI provider connections (`Anthropic`, `OpenAI`, `Google`, `AIGateway`) provide the same `GenerateText` and `GenerateObject` request types — the `connectionId` selects the provider. These make a single, one-shot model call and return the result. They can be used as page requests, or as steps in [API endpoint routines](/api) — useful for classify, extract, summarize, or decide steps where the surrounding logic stays in your routine.
 
 For multi-step tool use, see [agents](/agents-introduction) instead.
 
-### Limits and cancellation
-
-Two limits bound every AI request. Set them on the request, or on the connection as the default for every request and agent that uses it:
-
-- `maxOutputTokens` caps the tokens one call generates. Without it, the provider's default applies.
-- `timeout` cancels a call, retries included, that runs longer than this many milliseconds. The request fails with a `ServiceError`.
-
-A call is also cancelled when the request that started it closes, so a provider call that nobody is waiting for stops instead of running, and billing, to its end:
-
-- A page request, or an API endpoint called over HTTP, is cancelled when the server's request timeout (`config.requestTimeout`, default 30 seconds) answers first, which fails it with a `ServiceError`. On the Node server it is also cancelled when the client disconnects; that is logged as a warning, not an error.
-- On [Vercel](/vercel), only the request timeout cancels a call: a client disconnect does not, and the call runs until it finishes, times out, or the function reaches `config.vercel.maxDuration`. Keep `config.requestTimeout` below `maxDuration`, or Vercel stops the function before the timeout can cancel anything (the build warns when it is not).
-- Endpoints with `async: true`, `detached: true` calls, and scheduled runs answer at once and then run to their end: the closing request does not cancel them.
-- An agent chat turn runs to its end after the client disconnects, so its `onFinish` hooks still save the conversation. Bound it with the agent's `timeout` and `maxSteps`. A `CallAgent` step is cancelled with the routine that runs it.
-
-When a provider rate-limits a call (HTTP 429), the model client retries it (`maxRetries`), waiting as long as the provider's `Retry-After` asks, up to a minute. If the retries run out, the request fails with a `ServiceError`, as a 429 does on every connection, and the server log records the `Retry-After` value.
-
 ### GenerateText
 
 Generates text from a prompt.
@@ -125,9 +106,7 @@ Generates text from a prompt.
 - `prompt: string`: Text prompt. Use either `prompt` or `messages`, not both.
 - `messages: object[]`: Model messages (`{ role, content }`). Use either `prompt` or `messages`, not both.
 - `system: string`: System prompt.
-- `allowSystemInMessages: boolean`: Default: `false` - Allow `system` role messages in `messages`. A system message instructs the model as the app itself, so without this a request whose `messages` include one fails. Only set it when the messages come from the app, never from a user (for example a message list built from `_payload`). Use `system` for the system prompt.
-- `maxOutputTokens: number`: Maximum number of tokens to generate. Defaults to the connection's `maxOutputTokens`.
-- `timeout: number`: Milliseconds the model call may take, retries included, before it is cancelled. Defaults to the connection's `timeout`.
+- `maxOutputTokens: number`: Maximum number of tokens to generate.
 - `temperature: number`: Sampling temperature (0 to 2).
 - `topP: number`: Nucleus sampling.
 - `topK: number`: Only sample from the top K options for each subsequent token.
@@ -251,142 +230,3 @@ api:
                 category:
                   _step: classify.object.category
 ```
-
-### Decide
-
-Makes typed decisions about a piece of state: pick one of a set of options, judge a statement true or false, or place something on an ordered scale. Every answer comes back in the shape you declared, with a confidence, so a routine can branch on it — and send a low-confidence answer to a person instead.
-
-The build knows every option, so a routine branch that compares an answer with a name that is not one of its options — `_eq: [{ _step: triage.team.choice }, bil-ling]` — fails the build rather than silently never matching. A `_step` reference to a question the step does not ask, or to a field its answer does not have, fails the build the same way.
-
-`Decide` has two backends behind the same interface:
-
-- `evaluation` — an evaluation model, such as TypeSafe's Jev (`typesafe-ai/jev`) on the [AI Gateway](/AIGateway). It returns a probability for every option in one pass, in under a second, and bills input tokens only. The default, and only available, on the `AIGateway` connection.
-- `structured-output` — any language model, asked for the answers as structured output that follows a schema of the questions. Providers do not all enforce the schema, so Lowdefy checks each answer against its question (see `null` answers below). Available on every AI connection, and the default on `Anthropic`, `OpenAI` and `Google`. Its confidence is the model's own estimate, not a calibrated probability.
-
-Keep the state to what the questions need: an evaluation model reads the state literally, and accuracy drops as it grows. A `Decide` is never a security boundary — text in the state can argue for its own answer.
-
-#### Properties
-
-- `model: string`: **Required** - Model id to decide with, e.g. `typesafe-ai/jev`. Pin a version wherever a confidence threshold is tuned — a `latest` alias changes answers on release.
-- `state: string | object | array`: **Required** - What the questions are about.
-- `questions: object`: **Required** - Named questions, answered together against the state. Each question id becomes a key of the response (`usage` is reserved). A question is one of:
-  - `{ choice: string, options: object }` - Pick exactly one option. `options` maps option names (2 to 255) to what each one means (`null` for no description).
-  - `{ yesno: string, criteria?: { yes: string, no: string } }` - Judge a statement true or false. `criteria` optionally says what counts as yes and as no.
-  - `{ score: string, levels: string[] }` - Place the state on 2 to 10 ordered levels, lowest first.
-- `backend: enum`: `evaluation` or `structured-output`. Defaults to `evaluation` on `AIGateway` and `structured-output` elsewhere.
-- `maxOutputTokens: number`: Maximum number of tokens to generate, on the `structured-output` backend (an evaluation model generates no text). Defaults to the connection's `maxOutputTokens`.
-- `timeout: number`: Milliseconds the model call may take, retries included, before it is cancelled. Defaults to the connection's `timeout`.
-- `maxRetries: number`: Maximum number of retries. Defaults to 2.
-- `providerOptions: object`: Provider-specific options, keyed by provider (e.g. `gateway: { zeroDataRetention: true }`).
-
-#### Response
-
-One key per question, plus `usage`:
-
-```yaml
-<choice question>:
-  choice: string # The chosen option.
-  confidence: number # 0 to 1.
-  probabilities: object # Probability per option (evaluation backend), else null.
-<yesno question>:
-  answer: boolean # true when the statement holds.
-  probability: number # Probability that it holds.
-  confidence: number # Probability of the given answer.
-<score question>:
-  level: string # The most likely level.
-  index: number # Its position, from 0.
-  score: number # Fractional position on the scale (evaluation backend).
-  confidence: number # 0 to 1.
-  probabilities: object # Probability per level (evaluation backend), else null.
-usage: object # Token usage ({ inputTokens, outputTokens, totalTokens }).
-```
-
-An answer outside its question's terms (an option or level the question does not have, a yes/no that is not true or false, or no answer at all) comes back with every field `null`, its confidence included. `_lt` compares `null` below any number, so a confidence gate such as `_lt: [{ _step: triage.team.confidence }, 0.7]` holds it back like a low-confidence answer, and a branch on the choice matches none of the options.
-
-#### Examples
-
-###### Route a support ticket, and queue uncertain ones for a person:
-
-```yaml
-api:
-  - id: triage-ticket
-    type: Api
-    routine:
-      - id: triage
-        type: Decide
-        connectionId: gateway
-        properties:
-          model: typesafe-ai/jev
-          state:
-            subject:
-              _payload: subject
-            body:
-              _payload: body
-          questions:
-            team:
-              choice: Which team handles this ticket
-              options:
-                billing: Invoices, charges and refunds
-                tech: Outages and bugs
-                sales: Plans and quotes
-            down:
-              yesno: The customer says the service is down
-          providerOptions:
-            gateway:
-              zeroDataRetention: true
-      - ':if':
-          _lt:
-            - _step: triage.team.confidence
-            - 0.7
-        ':then':
-          - id: queue_review
-            type: MongoDBInsertOne
-            connectionId: review_queue
-            properties:
-              doc:
-                subject:
-                  _payload: subject
-                suggested_team:
-                  _step: triage.team.choice
-          - ':return':
-              status: review
-      - ':if':
-          _eq:
-            - _step: triage.team.choice
-            - billing
-        ':then':
-          - id: open_billing_case
-            type: CallApi
-            properties:
-              endpointId: open-billing-case
-      - ':return':
-          team:
-            _step: triage.team.choice
-          urgent:
-            _step: triage.down.answer
-```
-
-###### The same decision with any language model:
-
-```yaml
-requests:
-  - id: rate_review
-    type: Decide
-    connectionId: claude
-    payload:
-      review:
-        _state: review_text
-    properties:
-      model: claude-haiku-4-5
-      state:
-        _payload: review
-      questions:
-        sentiment:
-          score: How positive is this review
-          levels:
-            - Very negative
-            - Negative
-            - Neutral
-            - Positive
-            - Very positive
-```

package/content/connections/awss3.md

@@ -70,7 +70,6 @@ The same connection and requests work against any S3-compatible service — set
 
 Request types:
   - AwsS3GetObject
-  - AwsS3HeadObject
   - AwsS3PresignedGetObject
   - AwsS3PresignedPostPolicy
   - AwsS3PutObject
@@ -197,59 +196,6 @@ api:
             _step: put_copy.key
 ```
 
-### AwsS3HeadObject
-
-The `AwsS3HeadObject` request checks on the server whether an object exists in the bucket and
-returns its metadata without reading its content. Use it in an [API endpoint](/lowdefy-api) routine
-to confirm a file uploaded from the browser with an `AwsS3PresignedPostPolicy` actually arrived, and
-to take its size and content type from S3 rather than from the caller. The connection must allow
-reads (`read` is `true` by default).
-
-#### Properties
-- `key: string`: __Required__ - Key under which the object is stored.
-- `versionId: string`: VersionId used to reference a specific version of the object.
-
-Returns `{ exists: true, bucket, key, size, contentType, etag, lastModified, versionId }` for an
-object that exists (`size` in bytes, `lastModified` a date, `contentType` and `versionId` only when
-S3 returns them), and `{ exists: false, bucket, key }` when S3 answers not found. Any other S3 error
-throws. S3 answers not found only when the credentials may list the bucket (`s3:ListBucket`);
-without that permission a missing object is a 403 error, which throws.
-
-#### Examples
-
-###### Record an upload only once it is in the bucket:
-```yaml
-api:
-  - id: record_upload
-    type: Api
-    routine:
-      - id: head_object
-        type: AwsS3HeadObject
-        connectionId: my_bucket
-        properties:
-          key:
-            _payload: key
-      - :if:
-          _not:
-            _step: head_object.exists
-        :then:
-          - :reject: The upload did not arrive.
-      - id: insert_file
-        type: MongoDBInsertOne
-        connectionId: files
-        properties:
-          doc:
-            key:
-              _step: head_object.key
-            size:
-              _step: head_object.size
-            contentType:
-              _step: head_object.contentType
-      - ':return':
-          size:
-            _step: head_object.size
-```
-
 ### AwsS3PutObject
 
 The `AwsS3PutObject` request writes an object to the bucket from the server. It accepts base64 encoded