AI assistants are good at writing captions, but until recently they couldn’t see your calendar, your inbox or last month’s numbers — so you copied and pasted between tabs. An MCP server for social media changes that: it lets an assistant you already use work directly with your social tools, within limits you set. Here’s what that means in practice, what it’s useful for, and the guardrails that make it safe for a team.
What is the Model Context Protocol (MCP)?
The Model Context Protocol is an open-source standard for connecting AI applications to external systems. Anthropic introduced it in November 2024, and it is now supported by a range of AI assistants and developer tools, including Claude and ChatGPT. The official site describes it as being “like a USB-C port for AI applications”: one standard plug instead of a custom integration for every pairing.
There are three parts to know, as set out in the MCP architecture overview:
- The host is the AI application you use — for example a desktop assistant.
- The client is the connector inside that app that talks to one server.
- The server is a program that exposes data and actions to the assistant.
Servers mainly offer tools (actions the assistant can take, such as “create a draft post”), resources (data it can read) and prompts (reusable templates). When you connect a server, your assistant can see which tools exist and decide when to use them — and you can see what it did.
What an MCP server for social media lets your assistant do
For a social media team, an MCP server turns your assistant from a writing helper into a teammate that can work with real accounts, real calendars and real numbers. Open Motives’ MCP server offers 42 tools across three broad jobs.
Draft and schedule posts
Give your assistant a brief, and it can write drafts per network, place them in your calendar and pick times — without you copying text between windows. For example:
“Plan next week’s posts from this brief: autumn menu launch, two Reels, one LinkedIn hiring post. Use our best times.”
Work the inbox
Your assistant can help triage and draft replies across your connected conversation channels — useful for summarising a busy morning or preparing answers to common questions. It works across the conversation channels you’ve connected to the workspace.
“Summarise unanswered messages from today and draft replies to the delivery questions.”
Read analytics and answer questions
Instead of exporting a report and pasting it into a chat, ask directly:
“How did last month go? Which posts drove the most engagement, and what should we do more of?”
The assistant reads your analytics through the server and answers from your actual numbers, which you can check in the app.
Example: a week of posts from one brief
Here’s how this looks for Kettle & Co., the bakery we use as a running example.
The owner opens their AI assistant, which is connected to the Open Motives MCP server, and pastes a short brief: cardamom buns are back from Thursday, autumn menu launches this week, Riverside deliveries now run on Saturdays.
The assistant drafts captions for each network, applies the workspace’s best times, and schedules them. The result in the app: five drafts scheduled, best times applied, awaiting approval. Nothing goes live until a teammate reviews the drafts in the calendar, because the workspace’s approval rules still apply.
Later that week, the owner asks the same assistant how the launch is going. It reads the analytics for the workspace and answers with the numbers behind its summary, which the owner can open in the app to check.
That’s the shape of good AI help: it does the legwork, and a person makes the call.
The guardrails that make it safe for a team
Connecting an AI assistant to accounts that speak for your brand needs care. The MCP project’s own security best practices recommend a least-privilege approach — give a connection only the access it needs. Here’s how that translates for social media.
Scoped keys
An assistant should never get the keys to everything. In Open Motives, Agent API and MCP keys are scoped to one workspace and the accounts you pick. An agency can give an assistant access to one client’s Instagram and Facebook without exposing any other client. API keys are stored hashed, and channel credentials are encrypted at rest with AES-256-GCM.
Approvals stay in place
Posts created through MCP follow the same approval rules as posts your team writes. If a workspace requires internal or client approval, AI-created drafts wait in the calendar like any other post. That means you can let an assistant draft freely without changing how content gets signed off.
An audit log you can check
You should always be able to answer “who did what, and when?” — whether the “who” was a person or an assistant. Open Motives keeps an activity log of who did what, when and from where, with filters and CSV export.
The rest of your security settings
The usual protections still apply: two-factor sign-in (which organisations can require for everyone), tenant isolation at the database layer and permission checks on every API call. See the security page for details.
MCP vs REST API vs built-in AI
Open Motives offers three ways to bring AI and automation into your work. They suit different people:
| Option | Best for | How you use it |
|---|---|---|
| MCP server | People who already work in an AI assistant | Ask in plain language; the assistant calls the tools |
REST API (/api/v1/) | Developers building their own integrations | Write code against documented endpoints |
| Built-in AI (AI Employee, Copilot) | Teams working inside Open Motives | AI features inside the inbox and composer |
You don’t have to choose just one. Many teams use built-in AI day to day and an assistant over MCP for planning and reporting. The Agent API and MCP feature page covers both developer options.
How to get started
- Pick one job first. Reporting questions are a good start, because they only read data.
- Create a key for one workspace and choose only the accounts the assistant needs.
- Add the server to your MCP-capable assistant, following your assistant’s instructions for adding a connector. Setup details are in the developer docs.
- Ask read-only questions for a week. Check the answers against the app to build trust.
- Move on to drafts. Let the assistant plan posts, with approvals switched on for that workspace.
- Review the activity log regularly, at least while you’re getting started.
Things to watch
- AI can be wrong. Treat its drafts and summaries as a first pass. That’s what approvals are for.
- Inbound messages are untrusted text. Customer messages may contain instructions aimed at an AI. Keep human review on anything that publishes or replies in your brand’s name.
- Costs sit with your assistant. Your AI assistant’s usage is billed by whoever provides it, separately from Open Motives.
- Keep scopes tight. Create separate keys per workspace and per purpose, and remove ones you no longer use.
Frequently asked questions
What is an MCP server?
An MCP server is a program that exposes data and actions to AI assistants using the Model Context Protocol, an open standard. Once connected, an assistant can use the server’s tools on your behalf.
Which AI assistants can connect to Open Motives?
Any AI application that supports MCP clients can connect, including assistants such as Claude and ChatGPT where MCP connections are available. Check your assistant’s documentation for how to add a server.
Can the AI publish posts without approval?
Posts created through MCP follow your workspace’s approval rules. If a workspace requires approval, AI-created drafts wait for a teammate just like any other post.
Is MCP access included in the Free plan?
Yes. Both Free and Pro include the Agent API and MCP server; plan limits such as connected accounts still apply.
Conclusion
An MCP server lets the AI assistant you already use plan posts, help with the inbox and answer questions about your results — without copy and paste. The value comes from pairing that with the right limits: keys scoped to one workspace, approvals that still apply, and a log of everything that happened.
Curious how it would fit your team? Get in touch or compare plans.