How do you automate social media posts with n8n?

What n8n does in this job

n8n presents itself as workflow automation: a workflow starts from a trigger and passes data from node to node. Its integrations pages list the pieces this job needs: a Schedule Trigger, an HTTP Request node, a LinkedIn integration and an MCP Client Tool. None of them publishes by itself. Posting is still done by an API that the workflow calls, so what decides whether it works is that API: its sign-in, its limits and its rules.

Calling the networks directly

Read on 10 October 2026. Each network is a separate integration, with its own account type and its own sign-in:

• Instagram. Publishing through Meta’s API is two requests: create a media container, then publish it. A video container has to finish processing before it is published, so the workflow checks its status. The account must be a professional account, and an Instagram account is limited to 100 API-published posts in a 24-hour moving period. The page describes publishing at once and no publish time, so scheduling is the workflow’s job. n8n’s template library has community workflows that do exactly this with Graph API requests, such as one that creates the containers, waits for processing and publishes.
• LinkedIn. A member authorizes your app with OAuth 2.0 and the w_member_social permission, which LinkedIn describes as required to create a post on that member’s behalf, and the app then posts to LinkedIn’s API. A picture or video is registered and uploaded first. The page lists a limit of 150 requests per member a day. LinkedIn’s Posts API replaces the older ugcPosts API, and its API Terms also carry a line on automated posting, quoted in is social media automation allowed.
• TikTok. An app has to be audited before its posts can be public. Until then they are private, for at most five users in 24 hours, and TikTok wants the user to have expressly consented to the upload and to choose the privacy setting themselves.

Three networks means three credentials, three payload shapes and three sets of errors to branch on. That is the case for posting through one service.

Through Mellow Hub’s REST API from an HTTP Request node

Hub publishes to Instagram, TikTok, YouTube, X, LinkedIn, Threads, Bluesky, Pinterest and Facebook from one request. The sequence, with the address and fields as Hub’s API defines them:

1. Make a key. In Hub’s Agents screen, create an API key and choose the channels it may use, review or autopilot, a daily limit and an expiry. It is shown once. Keep it in an n8n credential, not in the workflow.
2. Send it as a header. Every request carries Authorization: Bearer <key> to https://www.mellow.world.
3. Read the channels once. GET /api/hub/v1/channels returns each connected account with its ID, which begins spc_. Put those IDs in the workflow.
4. Validate. POST /api/hub/v1/validate with the post publishes nothing and returns ok, issues and notes. It answers HTTP 200 even when ok is false, so branch on ok, not on the status code.
5. Create. POST /api/hub/v1/posts with the same body and an Idempotency-Key header. The key must be 8 to 120 characters of letters, digits and _ . : -, and it should come from the item, such as a row ID plus the date, not from a random number. Run the node again for the same row and Hub returns the original post; send different content under the same key and it is refused with idempotency_key_reused.
6. Read the result. GET /api/hub/v1/posts/<id> returns each network’s status and link. A post to several networks can be partial, with one network failed and the others published, so read the targets and do not simply send it all again.

A body, in short form: {"caption":"Our autumn menu starts today","channels":["spc_instagram_example","spc_linkedin_example"],"media":["https://example.com/menu.jpg"],"scheduledAt":"2026-10-14T09:00:00+02:00"}. Leave scheduledAt out to publish at once. A network with a shorter limit can have its own caption in perChannel. The media URL must still be public when the network fetches it.

With a review key the post comes back as pending_review and waits for a person to approve it in Hub’s Posts screen, so a workflow can prepare posts and leave the decision to you. With an autopilot key it goes out at scheduledAt, inside the daily limit. Revoking the key in Agents takes effect on its next request.

Hub’s own n8n node: built, not published

Mellow built an n8n community node for this API, n8n-nodes-mellow-hub, and four templates: a webhook that validates, creates and answers; a content calendar in Google Sheets; a daily AI-written caption that Hub checks before it goes out; and RSS items to channels. Its operations are Who Am I, Get Many channels, three media operations, and Validate, Create, Get, Get Many, Reschedule and Cancel for posts.

What is and is not true of it, as of 10 October 2026: its build succeeded and its 18 tests passed on 3 October, and n8n’s own linter was clean. It has not run inside a live n8n, the templates have not been imported into n8n, and the package is not published to npm or to n8n, so you cannot install it. Until it is, the HTTP Request recipe above does the same job.

And n8n’s MCP node?

n8n lists an MCP Client Tool node, a way for an n8n AI agent to use tools from an MCP server, and Hub’s MCP server is at https://www.mellow.world/mcp. We have not connected the two, so we cannot say that it works. A fixed workflow that posts the same shape every time is better served by REST. MCP suits an AI agent that decides which tool to call: what an MCP server is.

Hub is paid: plans start at $19 a month (Solo), after a 7-day trial with a card on file that costs nothing if you cancel before it ends, and a monthly plan's first month is 20% off. Publishing is unlimited on every plan, under a fair-use ceiling that stops automated floods (900 publications a month on Solo). One publication is one post on one account, so a post sent to five networks counts as five. There are no automatic overage charges.

Mellow’s own agent sessions have created posts through Hub’s MCP endpoint and have moved posts and added channels through this REST API, all with owner-made keys. No n8n workflow has been run against it.