MCP server for social media posting
An MCP server for social media posting lets an MCP-compatible AI client call posting and scheduling tools through one hosted URL.
{
"mcpServers": {
"social-media": {
"url": "{{API_BASE_URL}}/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}{"social-media":{"url":"{{API_BASE_URL}}/mcp","headers":{"Authorization":"Bearer YOUR_API_KEY"}}}
This guide explains how to configure the Model Context Protocol connection with a Bearer token, which clients are named as compatible, and how to check the setup before asking an agent to publish or schedule content.
What this MCP server does
Model Context Protocol, commonly called MCP, is an open standard for connecting AI assistants to external tools and data sources. An MCP server for social media posting exposes social media capabilities as tools that a compatible client can discover and call, instead of requiring a separate integration for each assistant. The hosted setup described here uses one server URL and one credential, so the client configuration stays small and reviewable. The server is described as covering posting, scheduling, analytics, connections, profiles, and AI helpers, with 34 tools in total, although the individual tool names are not enumerated on this page.
Use cases for agent-driven social posting
Use cases include asking an assistant to draft and schedule a week of posts in one prompt, cross-posting an article to LinkedIn and Threads, building agentic workflows that draft, review, and publish content, and giving a custom AI agent direct access to social channels. These examples are useful because they separate three different tasks: generating copy, choosing a time and destination, and publishing or scheduling through the connected accounts. Before running any of them, confirm that the relevant social accounts are connected and that the credential in the Authorization header is the intended access key. A prompt that asks for a schedule should state the destination accounts, the desired dates and times, and whether the agent should publish immediately or schedule a post for later review.
Why use a hosted MCP server
The hosted option requires no local process. You do not run npx, Docker, or a local server for the hosted endpoint; you add the URL to your client and supply the credential. That reduces setup work, but it also means the client must support the right transport. The source names Streamable HTTP as the transport, so a client that only supports a different MCP transport will not connect correctly even if the URL and token are right. Treat transport support as a prerequisite check, not as an assumption.
Setup steps
First, obtain an access key from the dashboard. Second, add the MCP server URL to your client configuration and set the Authorization header to a Bearer token using that key. Third, restart or reload the client if it requires a restart to pick up new MCP servers, then confirm the server appears in the client's tool list before sending a posting prompt. The field map and client compatibility table below make each part of that configuration explicit. If the server does not appear, work through the failure matrix rather than changing the URL path or header name, because both are part of the exact contract shown in the configuration example.
Worked example: draft, schedule, and cross-post
A worked flow can start with drafting: ask the agent to write several short posts from a supplied topic, keeping each draft separate so you can review tone and length. Next, ask it to schedule a post for a specific date and time on a named account, and check the returned schedule or post listing before assuming the time was accepted. Finally, ask it to cross-post the same article summary to LinkedIn and Threads, then verify each destination separately, because a successful call for one account does not prove the other account accepted the content. Keep prompts concrete about destination, timing, and whether the action is draft, schedule, or publish, and review the agent's output before it acts on connected accounts.
Failure checks and troubleshooting
Troubleshooting should follow the order of the connection: URL, transport, header, credential, then account connection. A wrong URL path will prevent the client from reaching the MCP endpoint, while a non-Streamable HTTP client may fail even with the correct URL. A missing or misnamed Authorization header, or a token that is not formatted as a Bearer token, will fail authentication. If authentication succeeds but posting fails, check that the intended social account is connected and that the prompt names a valid destination. Record which check failed so the fix addresses the actual layer instead of changing unrelated settings.
| Field | Type | Required | Value or format |
|---|---|---|---|
| mcpServers | object | Yes | Top-level object containing one entry per MCP server |
| mcpServers.social-media | object | Yes | Server entry in the public example; source server key retained privately |
| mcpServers.social-media.url | string | Yes | {{API_BASE_URL}}/mcp |
| mcpServers.social-media.headers | object | Yes | Object containing request headers for the server |
| mcpServers.social-media.headers.Authorization | string | Yes | Bearer YOUR_API_KEY |
Field map derived from the exact client config shown on the source page; origin modularized as {{API_BASE_URL}} and server key neutralized in public copy.
| Client | Source statement | Transport check | Decision |
|---|---|---|---|
| Claude Desktop | Named as compatible | Streamable HTTP | Use the config example and confirm the server appears in the client |
| Claude Code | Named as compatible | Streamable HTTP | Use the config example and confirm the server appears in the client |
| Cursor | Named as compatible | Streamable HTTP | Use the config example and confirm the server appears in the client |
| ChatGPT | Named as compatible | Streamable HTTP | Use the config example and confirm the server appears in the client |
| Other MCP clients | Described as a growing list of tools and IDEs | Must support Streamable HTTP MCP transport | Check transport support before adding the URL |
Based only on client names and transport statements in the supplied source page.
| Failure | Likely cause | Check | Fix |
|---|---|---|---|
| Server does not appear in client | Wrong URL path or unsupported transport | Confirm url ends with /mcp and client supports Streamable HTTP | Correct the URL path and use a Streamable HTTP client |
| Authentication fails | Missing API key or token not formatted as Bearer token | Inspect headers.Authorization value | Set Authorization to Bearer followed by the access key |
| Authentication fails despite a key | Wrong header name | Confirm header is named Authorization | Rename the header to Authorization exactly |
| Agent cannot schedule a post | Account not connected or prompt lacks destination or time | Check connected accounts and prompt details | Connect the account and restate destination, date, and time |
Setup failures inferred from the exact config contract and transport requirement in the source page.
Frequently asked questions
What is MCP?
Model Context Protocol is an open standard for connecting AI assistants to external tools and data sources. In this setup, the social media API is exposed as MCP tools so an MCP-compatible client can call posting, scheduling, and related capabilities through the configured server URL.
Do I need to install anything?
No local installation is required for the hosted setup described here. You add the server URL to your AI client's MCP configuration, with no npm install, Docker, or local process to run for the hosted endpoint.
Which AI clients support MCP?
Claude Desktop, Claude Code, Cursor, and ChatGPT are named as compatible clients, along with other tools and IDEs that support Streamable HTTP MCP transport. Check that your client supports Streamable HTTP before adding the URL.
How does authentication work?
Pass your access key as a Bearer token in the Authorization header. The server forwards that credential for authentication, credit accounting, and rate limiting when it calls the underlying REST API.
Can my agent schedule posts too?
Yes. Posting and scheduling capabilities are exposed through MCP, so an agent can draft content, schedule a post, list posts, and manage related integrations through the same configured connection.
Is the source code open?
The source page states that the source code is open and that self-hosting is an option, while the hosted version covers most needs. The repository path from the source is retained only in private editorial notes and is not presented as a public link here.