picked 0.0.101-next.2 → 0.0.101

package/docs/reference/dashboard/workflow-templates.md

@@ -0,0 +1,182 @@
+---
+type: reference
+layout: docs
+---
+
+# Workflow templates
+
+A template is text with `{{ … }}` in it. Picked fills it with data from the run before the
+node uses it. A tool node can have a template in each of its fields. A condition node can
+have a template in a text value of its rule.
+
+## Data that a template reads
+
+| Root              | Data                                                                   | Example                          |
+| ----------------- | ---------------------------------------------------------------------- | -------------------------------- |
+| `trigger`         | The event or the action that started the run.                          | `trigger.action.responses.email` |
+| `nodes.<node-id>` | The output of the last visit to a tool node. Refer to [Loops](#loops). | `nodes.n2.leads[0].id`           |
+| `run`             | The run. Refer to [Variables](#variables).                             | `run.startedAt`                  |
+| `workflow`        | The workflow. Refer to [Variables](#variables).                        | `workflow.name`                  |
+| `current`         | The node that reads the data. Refer to [Variables](#variables).        | `current.visit`                  |
+
+A template writes an index as `[0]`, such as `leads[0].id`. The `var` operator of a rule
+reads the same paths, and writes an index as a segment, such as `leads.0.id`.
+
+In the editor, the **Insert reference** button of a field shows the data that the node can
+read. The **Previous steps** tab shows `trigger` and `nodes`. The **Variables** tab shows
+`run`, `workflow` and `current`.
+
+## Variables
+
+| Path                | Type          | Value                                                                                     |
+| ------------------- | ------------- | ----------------------------------------------------------------------------------------- |
+| `run.id`            | Text          | The ID of the run, such as `wkrun_…`. In a test run, the ID of the test run.              |
+| `run.startedAt`     | Text          | The time that Picked recorded the run. In a test run, the time that the test run started. |
+| `run.preview`       | True or false | True for a run of the draft from a preview of the workspace, and for a test run.          |
+| `workflow.id`       | Text          | The ID of the workflow.                                                                   |
+| `workflow.name`     | Text          | The name of the workflow.                                                                 |
+| `current.id`        | Text          | The ID of the node that reads the variable.                                               |
+| `current.startedAt` | Text          | The time that the node started. If the node starts again, the time of the last start.     |
+| `current.visit`     | Number        | The number of the visit to the node, from 1. On a loop, each visit adds 1.                |
+
+A time is ISO 8601 in UTC, such as `2026-10-12T21:30:00.000Z`. The `date` filter changes the
+format and the timezone.
+
+Picked sets `run.startedAt` and `run.preview` one time, when it records the run. All the
+nodes of the run read the same values.
+
+Picked gets the `current` values when the node starts, and records the result of the node:
+
+- When a run continues after a stop, Picked uses the recorded results of the nodes that ran.
+  It does not run these nodes again.
+- If a node stops before Picked records its result, Picked starts the node again. The node
+  gets a new `current.startedAt`.
+- Picked does not start a failed node again. The run continues on the error exit of the node,
+  or stops at the node. Refer to the
+  [Workflow runs reference](/reference/dashboard/workflow-runs#run-status).
+- When the `retry` argument of an **External API** node is on, the node can send its request
+  more than one time. Each request uses the values that the node got when it started.
+
+When you save a workflow, Picked refuses a path below `run`, `workflow` or `current` that is
+not in this table. It also refuses `run`, `workflow` or `current` without a path below it.
+The error shows the path.
+
+## Loops
+
+A run can visit a node more than one time, on a loop. The path `nodes.<node-id>` gives the output of
+the last visit to that node that has a result. This is a visit that succeeded, or a visit
+that failed with an error.
+
+A node can read a node that comes after it on the loop. On the first visit, that node did
+not run, so the path has no value. In a condition, the value is `null`. In the editor,
+**Insert reference** shows these nodes with `(from the previous pass)`.
+
+## Templates in a condition
+
+A text value in a rule can have a template. Picked fills the template before it checks the
+rule.
+
+| Value                                                    | Result                                            |
+| -------------------------------------------------------- | ------------------------------------------------- |
+| Only one `{{ … }}`, such as `"{{ nodes.n2.count }}"`     | The value with its type. A number stays a number. |
+| Text with a `{{ … }}` in it, such as `"id-{{ run.id }}"` | Text.                                             |
+| Only one `{{ … }}`, with no value                        | `null`.                                           |
+
+A value that is only one `{{ … }}` must give text, a number, `true`, `false` or `null`. If it
+gives a list or an object, the condition stops with the code `condition_failed`. Text with a
+`{{ … }}` in it always gives text, also when the `{{ … }}` gives a list.
+
+Picked fills all the templates of a rule before it checks the rule. This includes a template
+in a part of `if`, `and` or `or` that the rule does not use. A template that gives a list or
+an object stops the condition, also when the rule does not use that part.
+
+A template in the rule of `map`, `filter`, `reduce`, `all`, `some` or `none` reads the data of
+the run. Picked fills it one time, before it checks the rule. It does not read the item.
+
+In a rule, only `{{` starts a template. A text value with only `{% … %}` in it stays as you
+write it. In a field of a tool node, `{% … %}` also starts a template.
+
+These values cannot have a template. Picked refuses the workflow when you save it:
+
+- The path of `var`.
+- The pattern and the flags of `regexp_matches`.
+- The message of `throw`.
+
+When you save a workflow, Picked checks each path in a template of a rule. It checks them as
+it checks a template in a tool node.
+
+## The `sha256` filter
+
+The `sha256` filter changes a value to its SHA-256 hash. The hash has 64 characters. Each
+character is a digit or a lowercase letter from `a` to `f`. You can use the filter in a tool
+node and in a rule.
+
+| Value                          | Result                                                                    |
+| ------------------------------ | ------------------------------------------------------------------------- |
+| Text                           | The hash of the text in UTF-8. Picked does not remove spaces in the text. |
+| A number, `true` or `false`    | The hash of the value as text. The number `42` gives the hash of `"42"`.  |
+| No value, `null` or empty text | Empty text.                                                               |
+| A list or an object            | The template stops.                                                       |
+
+The filter does not take an argument. A template that gives an argument to the filter stops.
+
+When a template stops, its node stops:
+
+- A tool node stops with a message that ends in `a template could not be rendered`. The node
+  does not use its error exit.
+- A condition stops with the code `condition_failed`.
+
+Picked does not have the `hmac_sha256` filter. When you save a workflow, Picked refuses a
+template that uses it.
+
+### A hash as a label key
+
+The namespace and the key of a label row can have a template. A hash fits the format of a
+label key. A namespace or a key has 1 to 64 characters, and a hash has 64 characters. A key
+that has a hash cannot have other text. Put other text in the namespace, for example
+`cms-lock-email`.
+
+When you save a workflow, Picked does not check the format of a namespace or a key that has a
+template. Picked checks the result when the node runs. If the result is not a correct namespace
+or key, the node stops. The node does not use its error exit.
+
+An empty key is not correct. Put the `default` filter before `sha256` when the template can
+give no value:
+
+```
+{{ trigger.action.responses.email | default: run.id | sha256 }}
+```
+
+## Example: a rule for a time of day
+
+This rule is true on days 1 to 5 of the week, from 08:00 to 16:59, in the timezone
+`Australia/Sydney`:
+
+```json
+{
+  "and": [
+    { "<=": ["{{ current.startedAt | date: '%u', 'Australia/Sydney' | plus: 0 }}", 5] },
+    { ">=": ["{{ current.startedAt | date: '%H', 'Australia/Sydney' | plus: 0 }}", 8] },
+    { "<": ["{{ current.startedAt | date: '%H', 'Australia/Sydney' | plus: 0 }}", 17] }
+  ]
+}
+```
+
+- `%u` gives the ISO day of the week, from 1 to 7.
+- `%H` gives the hour, from `00` to `23`.
+- `plus: 0` changes the text to a number.
+
+The offset of `Australia/Sydney` from UTC changes two times each year. The `date` filter
+follows the offset that applies on the date of the time.
+
+| Node started at, in UTC | Time in `Australia/Sydney`    | Branch  |
+| ----------------------- | ----------------------------- | ------- |
+| `2026-10-12T21:30:00Z`  | Day 2, 08:30, offset `+11:00` | `true`  |
+| `2026-06-12T21:30:00Z`  | Day 6, 07:30, offset `+10:00` | `false` |
+
+## Related
+
+- [Workflow test runs](/reference/dashboard/workflow-test-runs) — run the draft of a
+  workflow one time, and see the values that each node read.
+- [How to call an external API from a workflow](/how-to/call-an-external-api-from-a-workflow)
+  — read the answer of a call in a later node.

package/docs/reference/dashboard/workflow-test-runs.md

@@ -0,0 +1,85 @@
+---
+type: reference
+layout: docs
+---
+
+# Workflow test runs
+
+A test run runs the draft of a workflow one time, with the trigger input that you give.
+It shows what each node did. It saves nothing, so you can test a change before you
+publish it.
+
+## Where to start a test run
+
+| Page     | Address                       | Content                                                                         |
+| -------- | ----------------------------- | ------------------------------------------------------------------------------- |
+| Test run | `…/workflows/<workflow>/test` | The trigger form and the result. Open it with the **Test** tab of the workflow. |
+
+You must have the scope `ORGANIZATION_WORKSPACE_UPDATE` and each scope that the workflow
+uses. The page names each scope that you do not have.
+
+## Trigger input
+
+| Trigger | Input                           | Meaning                                                                                                                                            |
+| ------- | ------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------- |
+| Action  | One input for each action field | The answer to that field. Nodes read it as `trigger.action.responses.<key>`. If a field is empty and not required, the run sends no answer for it. |
+| Event   | **Entity ID**                   | The entity that the event is about. Nodes read it as `trigger.entityId`. An empty field uses the workspace.                                        |
+
+Each action field gets the input for its type:
+
+| Field                  | Input                                              |
+| ---------------------- | -------------------------------------------------- |
+| Text                   | A text box.                                        |
+| Number                 | A number box. The value must be a number.          |
+| True or false          | A checkbox.                                        |
+| Record                 | A box for JSON. The value must be valid JSON.      |
+| Field with options     | A list to choose one option.                       |
+| Field with many values | Checkboxes for options, or one value on each line. |
+
+A required field must have a value before the test run starts.
+
+## What runs
+
+| Node                                                                                         | In a test run                                                                                         |
+| -------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------- |
+| Tool that reads, such as **Search leads** or an **External API** `GET` or `HEAD`             | Runs.                                                                                                 |
+| Tool that writes, such as **Create lead**, **Set lead labels** or an **External API** `POST` | Does not run. The result shows the input that it would send.                                          |
+| **Set run labels** and **Remove run labels**                                                 | Do not run, because Picked does not keep a test run. The result shows the input that they would send. |
+| **Get run**                                                                                  | Runs, and gives the ID of the test run, which starts with `test_`, and no labels.                     |
+| Condition                                                                                    | Runs, and shows the branch that it took.                                                              |
+| Wait                                                                                         | Does not wait.                                                                                        |
+
+A later node that reads the output of a skipped node gets no value.
+
+A test run does not wait on a loop. It goes around the loop until a condition takes the
+exit, or until a node gets to the visit cap. Each visit shows as a node of the result, and
+a later visit shows its number, such as `visit 2`. Refer to the
+[workflow runs reference](/reference/dashboard/workflow-runs#visit-cap).
+
+## Result
+
+| Status      | Meaning                                          |
+| ----------- | ------------------------------------------------ |
+| `Succeeded` | The node did its work.                           |
+| `Skipped`   | The tool writes, so the test run did not run it. |
+| `Failed`    | The node failed, and shows the reason.           |
+
+When a node fails, the test run stops at that node and shows `Failed`. If an edge leaves
+the error exit of the node, the test run continues on that edge as a run does, and can
+show `Completed`. Refer to the [workflow runs reference](/reference/dashboard/workflow-runs#run-status).
+
+Each node shows its input and its output. A test run does not show in **Runs**, and it
+does not create or change leads or labels.
+
+## Run variables
+
+A test run gives its own values to `run.id`, `run.startedAt` and `run.preview`. Refer to
+[Workflow templates](/reference/dashboard/workflow-templates#variables).
+
+## API and MCP
+
+You can also do this task with the [Management API](/reference/management-api/schema), the [MCP server](/reference/tools/mcp), or code that you send to [`picked_run`](/reference/tools/mcp-run).
+
+| Task | GraphQL | MCP tool | Run |
+| --- | --- | --- | --- |
+| Start a test run | [`workflowTestRun`](/reference/management-api/schema/workflow#workflowtestrun) | [`test_run_workflow`](/reference/tools/mcp#test_run_workflow) | `tools.test_run_workflow({ workflow_id })` |

package/docs/reference/dashboard/workspace-versioning.mdx

@@ -0,0 +1,42 @@
+---
+type: reference
+layout: docs
+---
+
+import { Screenshot } from '../../../../components/Screenshot'
+
+# Workspace versioning
+
+When you publish, Picked makes an immutable snapshot of your workspace. A project
+always serves exactly one version, the *live* version. A roll-back moves the live
+pointer of the project to an earlier snapshot. It changes no data.
+
+<Screenshot
+  route="/{orgSlug}/{projectId}/publish-history"
+  alt="The Publish history page with published workspace versions and the Live and Restore actions"
+  hide={['[data-testid="version-row-relative-time"]']}
+/>
+
+## Publish history
+
+The **Publish history** page lists each published version of every workspace in the
+project, the newest first. The version that Picked serves now carries a **Live**
+badge.
+
+## API and MCP
+
+You can also do this task with the [Management API](/reference/management-api/schema), the [MCP server](/reference/tools/mcp), or code that you send to [`picked_run`](/reference/tools/mcp-run).
+
+| Task | GraphQL | MCP tool | Run |
+| --- | --- | --- | --- |
+| List the published versions | [`project`](/reference/management-api/schema/project#project) | [`list_workspace_versions`](/reference/tools/mcp#list_workspace_versions) | `tools.list_workspace_versions({ project_id })` |
+| Restore a version | [`workspaceVersionPublish`](/reference/management-api/schema/workspace#workspaceversionpublish) | [`guide_workspace_version_publish`](/reference/tools/mcp#guide_workspace_version_publish) | `tools.guide_workspace_version_publish({ workspace_version_id })` |
+
+The MCP server does not restore a version. `guide_workspace_version_publish` gives the
+link to the **Publish history** page in the dashboard.
+
+## Related
+
+- [About workspaces and versions](/explanation/workspaces-and-versions) — why a roll-back
+  changes no data.
+- [How to roll back a published version](/how-to/roll-back-a-published-version)

package/docs/reference/dashboard/workspaces.mdx

@@ -0,0 +1,91 @@
+---
+type: reference
+layout: docs
+---
+
+import { Screenshot } from '../../../../components/Screenshot'
+
+# Workspaces
+
+A workspace is one version of the configuration of your project. Use a workspace to hold
+a change, to see a preview of it, and to publish it when it is ready.
+
+<Screenshot
+  route="/{orgSlug}/{projectId}/workspaces"
+  alt="The workspaces list inside a project on the Picked dashboard"
+  hide={['[data-testid="workspace-row-relative-time"]']}
+/>
+
+## How workspaces work
+
+Each project has a workspace that holds your experiences, feature flags, steps, fields,
+actions, and conditions. When you make a change in the dashboard, that change stays in
+the workspace until you publish it. A change to a flow, to a feature flag, or to a step
+all behave the same way.
+
+A workspace is in one of three states:
+
+- **Live** — the workspace that the SDK serves to your visitors. Each project has
+  exactly one live workspace.
+- **Draft** — a workspace that no visitor sees. You can edit a draft, see a preview
+  of it, and publish it to make it live.
+- **Archived** — a draft that you put to one side. An archived workspace is still
+  editable, but it does not appear in the default workspaces list.
+
+## Change previews
+
+The publish modal shows a side-by-side change preview. It compares the current
+workspace against the last published snapshot. Use it to see exactly what your
+visitors get after you publish.
+
+<Screenshot
+  route="/{orgSlug}/{projectId}/workspaces/{workspaceId}/experiences?action=publish"
+  alt="The publish modal with a change preview of added experiences, steps, and conditions"
+  selector='[role="dialog"]'
+  wait='[role="dialog"]'
+  hide={['[data-testid="publish-diff-replaces-version"]']}
+  renderWidth={720}
+  renderHeight={579}
+/>
+
+The preview puts the changes into groups: experiences, feature flags, steps, fields,
+actions, and conditions. Each row shows whether the record was **added**, **removed**,
+or **modified**. For a modified record, expand the row to see which properties changed.
+
+Some changes break other records. An example is a field that you remove while other
+steps still point at it. The change preview shows a banner above the diff for these
+entries, so that you can look at them again before you publish.
+
+The first publish of a workspace has no earlier snapshot to compare against. The
+publish modal shows the contents of the workspace as the first publish, with no diff.
+
+## Unpublished changes
+
+The workspace records whether anything changed since the last publish. The toolbar then
+shows a warning badge. It counts a change to any of these:
+
+- An experience or a feature flag.
+- A step, a field, an action, or a condition.
+
+## API and MCP
+
+You can also do this task with the [Management API](/reference/management-api/schema), the [MCP server](/reference/tools/mcp), or code that you send to [`picked_run`](/reference/tools/mcp-run).
+
+| Task | GraphQL | MCP tool | Run |
+| --- | --- | --- | --- |
+| List the workspaces | [`project`](/reference/management-api/schema/project#project) | [`list_workspaces`](/reference/tools/mcp#list_workspaces) | `tools.list_workspaces({ project_id })` |
+| Open a preview of the workspace | [`workspace`](/reference/management-api/schema/workspace#workspace) | [`get_preview_url`](/reference/tools/mcp#get_preview_url) | `tools.get_preview_url({ workspace_id, target_url })` |
+| See the changes that you publish | [`workspaceDiff`](/reference/management-api/schema/workspace-version#workspacediff) | [`compare_workspace_versions`](/reference/tools/mcp#compare_workspace_versions) | `tools.compare_workspace_versions({ compare_workspace_id })` |
+| Publish the workspace | [`workspacePublish`](/reference/management-api/schema/workspace#workspacepublish) | [`guide_publish_workspace`](/reference/tools/mcp#guide_publish_workspace) | `tools.guide_publish_workspace({ workspace_id })` |
+| Find unpublished changes | [`workspace`](/reference/management-api/schema/workspace#workspace) | [`get_workspace_overview`](/reference/tools/mcp#get_workspace_overview) | `tools.get_workspace_overview({ workspace_id })` |
+
+The MCP server does not publish. `guide_publish_workspace` gives the link to the
+**Publish** button in the dashboard.
+
+## Related
+
+- [About workspaces and versions](/explanation/workspaces-and-versions) — why a draft and
+  a version are separate.
+- [How to see a preview of a workspace](/how-to/preview-a-workspace)
+- [How to publish a workspace](/how-to/publish-a-workspace)
+- [How to archive or delete a workspace](/how-to/archive-or-delete-a-workspace)

package/docs/reference/delivery-api/experience-content.md

@@ -0,0 +1,152 @@
+---
+type: reference
+layout: docs
+---
+
+# Experience Content Endpoint
+
+Fetch the content entries that back the steps of a single experience.
+
+## Request
+
+```
+GET /api/v1/projects/:projectId/experiences/:id/content
+```
+
+| Parameter | In | Description |
+| --- | --- | --- |
+| `projectId` | path | The project the experience belongs to. |
+| `id` | path | The experience to fetch content for. |
+| `preview_token` | query | Optional. With it, the endpoint returns preview content, which is unpublished. |
+
+## Response
+
+```json
+{
+  "experienceId": "exp_123",
+  "versionId": "ver_456",
+  "content": [
+    {
+      "id": "entry_789",
+      "contentModelKey": "page-hero",
+      "data": {
+        "title": "Spring sale",
+        "category": { "contentEntryId": "entry_42", "contentModelKey": "category" }
+      }
+    }
+  ],
+  "references": [
+    {
+      "id": "entry_42",
+      "contentModelKey": "category",
+      "data": { "name": "Outerwear" }
+    }
+  ]
+}
+```
+
+| Field | Type | Description |
+| --- | --- | --- |
+| `experienceId` | `string` | The experience the content belongs to. |
+| `versionId` | `string \| null` | The published version id, or `null` on preview. |
+| `content[]` | `array` | The entries that the steps of the experience point at. |
+| `content[].id` | `string` | The content entry id. |
+| `content[].contentModelKey` | `string \| null` | The Key of the Model, or `null` when it cannot be resolved. |
+| `content[].data` | `object` | The entry data. It matches the schema of the Model. |
+| `references[]` | `array` | Every entry that a `reference` field reaches, in the same blob shape as `content[]`. |
+
+The published response and the preview response hold the same fields, and the same
+values, for the same content.
+
+## A reference field points at a separate blob
+
+A [Reference field](/reference/dashboard/content/models#reference-fields) arrives as a
+pointer, and not as the entry:
+
+```json
+{ "contentEntryId": "entry_42", "contentModelKey": "category" }
+```
+
+A repeatable Reference field arrives as an array of such pointers.
+
+The entry itself arrives in `references[]` under its own `id`. Look the pointer up
+there. The SDK does this for you, through
+[`useReference` and `getReference`](/reference/sdk/content).
+
+`references[]` holds each entry one time, for any number of fields that point at it. It
+omits the entries that are in `content[]` already.
+
+### What `references[]` omits
+
+Picked resolves a chain of references to a maximum depth of 5. It counts the depth from
+the entries in `content[]`. An entry is absent from `references[]` when Picked cannot
+resolve it:
+
+| Cause | Result |
+| --- | --- |
+| The target was deleted | The pointer stays. The entry is absent. |
+| The target is in another workspace | The pointer stays. The entry is absent. |
+| The chain makes a cycle | The pointer stays. The entry is absent. |
+| The chain is deeper than 5 | The pointer stays, with its `contentModelKey`. The entry is absent. |
+
+A pointer with no entry in `references[]` means that there is no such entry. Picked sends
+no empty blob for it.
+
+:::warning
+A reference pointer also carries a `data` field with the content of the target inlined.
+This is for the SDK versions that came before `references[]`. Read `references[]`
+instead. A later release removes the inline copy.
+:::
+
+## An HTML asset comes back in the response
+
+A Model can have an [HTML asset](/reference/dashboard/content/asset-library) field. The
+value of that field arrives inside the response. The endpoint changes the reference to
+`{ id, content }`. There, `content` holds the HTML that you uploaded, as a string. It
+is `null` when the endpoint cannot resolve it:
+
+```json
+{
+  "id": "entry_789",
+  "contentModelKey": "page-hero",
+  "data": {
+    "markup": { "id": "asset_42", "content": "<section>...</section>" }
+  }
+}
+```
+
+The SDK renders the inlined HTML directly, with no extra storage fetch.
+
+## Status codes
+
+| Code | Meaning |
+| --- | --- |
+| `200` | Success. |
+| `401` | Invalid `preview_token`. |
+| `404` | The experience has no published content. |
+
+## Example
+
+```bash
+curl https://api.picked.app/api/v1/projects/proj_123/experiences/exp_123/content
+```
+
+```json
+{
+  "experienceId": "exp_123",
+  "versionId": "ver_456",
+  "content": [
+    { "id": "entry_789", "contentModelKey": "page-hero", "data": { "title": "Spring sale" } }
+  ],
+  "references": []
+}
+```
+
+## Related
+
+- [About the two APIs](/explanation/the-two-apis)
+- [About content references](/explanation/content-references) — why a reference is a pointer.
+- [Content](/reference/sdk/content) — how the SDK resolves a pointer.
+- [Project config](/reference/delivery-api/project-config) — the endpoint that links a step to an entry.
+- [Server Snapshot](/reference/sdk/ssr/serialize) — how the SDK reads this content.
+- [Preview, cache, and crawlers](/reference/delivery-api/preview-caching-crawling)

package/docs/reference/delivery-api/index.md

@@ -0,0 +1,45 @@
+---
+type: reference
+layout: docs
+---
+
+# Delivery API
+
+The Delivery API is a public REST API. It is read-only, and it serves the published
+configuration of your project together with your page content. It is separate from the
+GraphQL management API.
+
+## Base URL and versioning
+
+All Delivery API routes are versioned under `/api/v1`:
+
+```
+https://api.picked.app/api/v1
+```
+
+## Published content and preview content
+
+Each endpoint serves two surfaces:
+
+- **Published** content is anonymous. Anybody can read your published configuration and
+  your published content.
+- **Preview** content needs a `preview_token` query parameter. With it you can render
+  an unpublished draft before you make it live.
+
+The management API holds the preview token of a workspace in the `previewToken`
+field. The preview action in the dashboard also makes one for you.
+
+## Endpoints
+
+| Endpoint | Description |
+| --- | --- |
+| [Project Config](/reference/delivery-api/project-config) | `GET /api/v1/projects/:projectId` returns the published configuration of the project, with the content link of each step. |
+| [Experience Content](/reference/delivery-api/experience-content) | `GET /api/v1/projects/:projectId/experiences/:id/content` returns the Content Entries of an experience. |
+
+Both endpoints share some behaviour. See
+[Preview, cache, and crawlers](/reference/delivery-api/preview-caching-crawling).
+
+## Related
+
+- [About the two APIs](/explanation/the-two-apis) — which one to use, and why they are
+  separate.