
An automation that turns an approved article into a social draft has a clear job. It reads known fields, creates one publication, and stores the returned ID. You can add AI rewriting, media, and scheduling later without changing that basic contract.
Postbean’s REST API gives n8n or Zapier an HTTP destination for that workflow. This article describes a custom API connection, not a claim that Postbean has a dedicated native node or an installed Zapier app. The examples use verified Postbean routes.
Build the first version around drafts. A workflow test can run more than once, and its sample data may be older than you expect. Explicit draft state lets you inspect what was saved before adding any public send.
Define the event that should create content
Choose a source event with an editorial meaning. “Article approved for promotion” is easier to control than “a row changed.” A spelling correction in a spreadsheet should not create a second public announcement.
Give the event a stable identifier. Your automation should retain the source ID, destination workspace, selected account IDs, and eventual Postbean ID in a store you control. That store can be part of your existing application or workflow data design; it is not a Postbean feature being configured here.
Decide what happens when the source changes after draft creation. Usually the useful action is an update to the same post. If your workflow cannot distinguish a new article from a revised one, fix that before connecting it to publication.
Prepare one workspace credential
In Postbean, create an API key from Settings → Developers after reviewing and enabling developer access. Assign the key to the intended project. For this draft workflow, select accounts:read, posts:read, and drafts:write.
Keep the token in the automation tool’s credential storage or private configuration. Do not map it from an editable spreadsheet cell. A content editor should not have to see or retype the secret to change a caption.
Resolve the account once through GET https://postbean.app/api/social-accounts, then confirm its platform and username. Use that account’s actual numeric ID in subsequent requests. For multiple clients, use a deliberate credential mapping rather than a shared key and a client name embedded in the caption.
Configure the n8n request
n8n’s HTTP Request node supports setting a method, headers, and a JSON body. Add it after your source validation and duplicate check.
Set the method to POST and the URL to https://postbean.app/api/posts. Configure bearer authentication through a suitable credential, or a private header configuration, and send Accept: application/json plus Content-Type: application/json.
Map the approved source text into caption. Set post_type to text, draft to the boolean true, and social_account_ids to an array of selected numeric account IDs. Do not send the string "true" or turn the array into comma-separated text.
Run the node against a test article and inspect its response. Save the public post ID from the returned resource in your mapping store. Then read that post with GET /api/posts/{post} before marking the source event as successfully processed.
Configure the Zapier request
Webhooks by Zapier provides request actions. Its Custom Request option accepts raw JSON and custom headers, which is useful when preserving arrays and nested objects. Check your plan’s access to that action.
Choose POST, use the same Postbean URL, and configure the authorization, accept, and content-type headers. Put explicit JSON in the data field and map only the fields you intend to send. Validate the resulting JSON after inserting dynamic caption text, especially when it contains quotes or line breaks.
For either tool, the logical payload is:
{
"caption": "The approved article summary goes here.",
"post_type": "text",
"draft": true,
"social_account_ids": [42]
}
Replace 42 with an ID returned for your workspace. Use a destination that supports text posts. The sample is not suitable for a video-only destination simply because the request body is valid JSON.
Put duplicate protection before creation
Before POST /api/posts, look up the source event in your own store. If it already has a Postbean ID, read or update that post instead of creating another. If it is currently being processed, avoid starting a parallel creation attempt.
Store the returned ID as soon as the response is available. If the request times out, inspect the destination and your operation records before rerunning it. The API example does not promise an idempotency header, so do not assume adding an invented header will prevent duplicates.
This matters during development too. Running a node by hand and then running the whole workflow can process the same source twice. Label test content clearly and keep it in draft state while validating the mapping.
Make AI rewriting an inspectable step
If you add AI between the source and Postbean, preserve the original facts next to the generated caption in your own process. Ask the model to flag missing details instead of inventing prices or urgency.
Allow the reviewer to see both versions. A shorter caption may accidentally remove a condition such as “booking required,” while a more promotional one may add a claim nobody approved. Neither error is caught merely by checking whether the HTTP request succeeded.
Use per-network versions when needed. The REST field is platform_overrides, containing network-specific captions. Keep this mapping explicit so the compact version reaches the intended destination rather than replacing the shared text for every account.
Add media without passing local paths to a remote service
If the source includes a file, use Postbean’s upload flow: POST /api/media/upload-url, upload the bytes with PUT to the returned signed URL, and confirm through POST /api/media. The reservation includes the file name, MIME type, and byte size.
Do not send your bearer token to the signed upload URL. Attach the resulting media ID in the post’s complete media_ids list, and change post_type to match the content. A path on the automation worker’s disk is not a path on Postbean’s server.
Test file transfer separately before adding it to a batch. Keep failures attached to the source event so an operator can retry the appropriate stage without recreating the whole post.
Separate draft production from release
After review, a scheduling request should carry an explicit future time and the intended non-draft state. Display the workspace time zone when the reviewer chooses that time. An offset-free timestamp uses that zone; Z means UTC.
Immediate release of an existing post uses POST /api/posts/{post}/publish-now and posts:publish. Treat scheduling as a release decision too, rather than assuming a draft-writing key establishes a complete approval barrier.
Postbean’s Approvals screen is currently a coming-soon placeholder. If your workflow needs an approval step today, maintain it in your own process and require the release branch to check it. Do not describe that external decision as a native Postbean approval state.
Test failure handling before enabling the trigger
Exercise a missing credential, invalid account mapping, and caption that exceeds a target’s limit. Route each failure to a place someone actually checks. Content validation errors need editing; permission errors need access review; timeouts require checking whether the operation already happened.
After a scheduled send, inspect GET /api/posts/{post}/status and individual account outcomes. Saving a draft or receiving a queued status does not establish successful publication.
Enable the trigger only after one source event produces one readable draft and a repeated event reuses its ID. The developer guide provides the REST details. Keep the workflow’s operation record until you can trace each source item to the publication it created.
Sources
- n8n HTTP Request node — checked November 22, 2026
- Zapier webhook requests — checked November 22, 2026
- Postbean API guide — checked November 22, 2026