The MCP server for social media
Your AI can finally answer your customers.
Connect the assistant you already use to every social account you run. It reads your comments, schedules your posts, replies in your voice and reports what worked, within limits you set.
and any MCP client[ PICK YOUR ASSISTANT ]
Set up in two minutes, whichever one you use
Each guide shows the real screens and commands, checked against the vendor's own documentation, and says plainly what does not work yet.
Using Cursor, VS Code, Codex or your own code? The developer docs have the config for each.
[ WHAT IT DOES ]
Five jobs off your plate
Your AI does not get a vague licence to “manage your socials”. It gets these jobs, and nothing beyond them.
Tell you what needs an answer
It reads the comments and messages from every connected platform, already sorted by tone, and tells you which three matter today.
WHAT PEOPLE SAY
[ THE DIFFERENCE ]
Publishing is the easy half.
Plenty of tools let an AI schedule posts. Ask one what your customers said this morning, and you see the difference.
The conversation stops here.
Easy to check: ask your current tool the same question.
[ SETUP ]
Three steps, once
No SDK, no webhook, nothing to host.
- 01Create a key
In Echoia, open Settings, then Developers. Name the key and tick what it may do. Read only is a good first week.
- 02Paste one address
Add Echoia to your assistant like any MCP server, with the address below and your key. The guides above show every screen.
- 03Ask your first question
“Anything I should answer today?” It now sees your accounts, calendar and inbox, within exactly the limits of the key.
[ CONTROL ]
You decide how far it goes
Each permission is ticked separately when you create the key, and Echoia checks it on every request. Most people move down this list over a few weeks.
Changed your mind? Revoke the key in one click: it stops on the next request.
[ STRAIGHT ANSWERS ]
Before you connect anything
Which assistants work with Echoia?
Any client that speaks MCP, the open protocol most assistants now support. We keep step-by-step guides for Claude, ChatGPT, Gemini, Grok, OpenClaw and Hermes Agent, and the developer docs cover Cursor, VS Code and Codex.
Do I need to write code?
No. You create a key in Echoia and paste one address into your assistant's settings. Developers can also use the REST API directly.
Can it post an image I paste into the chat?
No, and no integration can. A pasted picture is something the model sees, not a file it can pass on. Give it a public image link, or use an agent that reads files, such as Claude Code.
What can an agent never do?
Connect or disconnect your social accounts, change your plan, touch your billing, or read a workspace you did not grant. Without the matching scope it cannot publish or reply either.
Does it cost extra?
No. The MCP server and the API are on every plan, the free one included. Each call counts against your plan like any other action, so an agent left running cannot quietly turn into an invoice.
What if it does something I did not want?
Revoke its key in Settings, Developers. It stops on the very next request, whatever it was doing.
Give your agent
something to work with.
Connect your accounts once. Your AI does the rest, within the limits you set. Free plan included.