# Post from Claude, Cursor or your terminal, not from another dashboard

> A hosted MCP server with 39 tools and a CLI, so you can draft, schedule and publish social posts from your AI assistant, your editor, or CI. $20/mo.

- Audience: Developers, indie hackers and technical founders who build in public
- Price: $20 per month, $192 per year, 14-day free trial.

## The situation

You build in public, which means every shipped thing is supposed to become a post. In practice the post never gets written, because writing it means leaving the editor and opening a dashboard designed for marketers.

You do not want a content calendar. You want the release you just tagged to become a thread, without a context switch.

Every other tool in this category treats the API as an afterthought behind an enterprise plan. The interface is the dashboard, and automation is something you are allowed to buy.

## What DunSocial does about it

### A hosted MCP server, included in the plan

Thirty-nine MCP tools across posts, drafts, memory, media and workspaces, at https://api.dunsocial.com/api/mcp over streamable HTTP. Not an add-on, not an enterprise tier: it is in the $20 plan.

### Works with the clients you already run

Claude Desktop, Claude Code, ChatGPT, Cursor, VS Code, Windsurf, Cline, Continue, and any other MCP-compatible client. Setup is a generated token and a Bearer header, about two minutes.

### A CLI for local agents and CI

The dunsocial CLI (`dunsocial` or `dun` on npm) is a thin API client with personal access tokens and device login, so a build step or a local agent can publish.

### Agent actions stay auditable

Posts created over MCP are tagged with an MCP source in the app, so you can always see what the agent did rather than guessing.

### Threads that survive being written by a machine

X threads of up to 25 posts, composed and published as one chain, with an exact preview before it goes out.

## How it works in practice

1. Generate a token: In the app: Settings, Developers, MCP. Or a personal access token for the CLI.
2. Point your client at the server: Streamable HTTP to https://api.dunsocial.com/api/mcp with an `Authorization: Bearer <token>` header.
3. Talk to it like a tool, because it is one: "Draft a launch thread from this changelog and schedule it for peak time." The assistant calls the tools, saves the draft, and schedules it.
4. Review before it ships: Drafts land in the app with exact per-platform previews. Approval is still human; publishing is automatic once it is scheduled.

## When DunSocial is the wrong choice

- You want a self-hosted deployment. DunSocial is a hosted product with a hosted MCP server.
- You want the model to generate images for you. It does not; upload media to the gallery.
- You want webhooks for inbound engagement. The surface is publishing, not listening.
- You need TikTok or Facebook. Neither is supported.

## Frequently asked questions

### What is the MCP server, in practice?

It exposes DunSocial as tools an AI agent can call: 39 of them, covering posts (list, get, schedule, publish now, threads, reschedule, cancel, delete), drafts, memory, media and workspaces. Any MCP-compatible client can drive the whole product in natural language.

### Is MCP access an extra cost?

No. It is included in the single $20 per month plan, along with every other feature.

### Can I publish from CI?

Yes, through the CLI. `dunsocial` (also aliased `dun`) is published on npm and authenticates with a personal access token or device login, which makes it usable from a pipeline as well as from a local agent.

### How long does setup actually take?

About two minutes: generate a token in Settings, Developers, MCP, then point your MCP client at https://api.dunsocial.com/api/mcp over streamable HTTP with the token as a Bearer header.

### Can an agent publish without me approving it?

Content the AI writes is reviewed and approved by a human before it goes live. An agent can draft and schedule; the approval step stays yours.

## About this document

This is the Markdown representation of https://www.dunsocial.com/for/developers. The HTML version of the same page is at the same URL. You can also request Markdown from any page by sending `Accept: text/markdown`.

Machine-readable summary of the whole site: https://www.dunsocial.com/llms.txt

## Company

- Product: DunSocial, https://www.dunsocial.com
- Legal entity: THISUX PRIVATE LIMITED, Chennai, Tamil Nadu, India
- Support: support@dunsocial.com
- Open the app: https://app.dunsocial.com/login