About xpost
xpost is an open-source agent skill and hosted service for publishing social posts across nine networks. It is for people using AI agents to manage social accounts, with optional human approval before posts are sent.
Key features
- 01Publish to nine social networks
- 02Require approval before publishing
- 03List accounts and posting rules
- 04Review post status and delivery receipts
- 05Install as a skill, plugin, or CLI
Use cases
- 01Publish one post to multiple accounts
- 02Hold posts for human approval
- 03Check per-account delivery results
- 04Manage social posts from an MCP client
MCP support
The hosted MCP server at `https://xpost.to/api/mcp` exposes social-account and posting capabilities to MCP clients. The npm package also provides a local MCP server.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://xpost.to/api/mcpclaude mcp add --transport http to-xpost-xpost https://xpost.to/api/mcp{"mcpServers":{"to-xpost-xpost":{"type":"http","url":"https://xpost.to/api/mcp"}}}Open in Cursor{"servers":{"to-xpost-xpost":{"type":"http","url":"https://xpost.to/api/mcp"}}}Open in VS Code{"mcpServers":{"to-xpost-xpost":{"type":"http","url":"https://xpost.to/api/mcp"}}}{"mcpServers":{"to-xpost-xpost":{"serverUrl":"https://xpost.to/api/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project README
xpost for your assistant
Post all your social media with your AI agent: X, Instagram, LinkedIn, Facebook, TikTok, YouTube, Threads, Bluesky and Pinterest, from Claude, ChatGPT, Cursor, Gemini CLI or any MCP client. Want a say first? Switch on approvals and nothing goes out until a person says yes.
This repo is the open-source skill (the workflow an agent should
follow) plus the plugin manifests that install it together with the hosted
MCP server at https://xpost.to/api/mcp. You sign in once in a browser
window; nothing to paste.
You need an xpost account — the first post is free, no card: https://xpost.to/signup
Install
Any agent — as a skill
npx skills add xpost-to/xpost-agent
This installs the skill only. Add the MCP server in your client as below,
or let the skill's REST fallback use XPOST_API_KEY.
Claude Code — plugin (skill + MCP server)
/plugin marketplace add xpost-to/xpost-agent
/plugin install xpost@xpost-agent
Nothing prompts you to sign in: type /mcp, choose xpost, then
Authenticate — your browser opens on the sign-in page. Or without the plugin:
claude mcp add --transport http xpost https://xpost.to/api/mcp
Gemini CLI — extension (skill + MCP server)
gemini extensions install https://github.com/xpost-to/xpost-agent
Gemini opens a browser to sign in to xpost on first use; /mcp auth xpost
signs in again.
Cursor
One click: Add xpost to Cursor
Or add it by hand in Settings → Tools & Integrations → New MCP Server:
{ "mcpServers": { "xpost": { "type": "http", "url": "https://xpost.to/api/mcp" } } }
Then npx skills add xpost-to/xpost-agent for the skill. The repo also
carries a Cursor plugin manifest (.cursor-plugin/plugin.json).
Grok Build
The repo carries a Grok Build plugin manifest (.grok-plugin/plugin.json,
skill + hosted MCP server) and its own marketplace catalog
(.grok-plugin/marketplace.json). Inside Grok Build, /marketplace browses
the catalog; from a shell, grok plugin install xpost --trust once the
source is known to it. Not yet in the official xAI index.
Muse Code (Meta)
Add the hosted server to ~/.config/muse/settings.json; Muse Code asks you
to sign in on the first tool call:
{
"schema_version": 1,
"mcp_servers": {
"xpost": { "transport": "streamable_http", "url": "https://xpost.to/api/mcp" }
}
}
Grok, Muse and other chat assistants with no MCP settings
Paste this into the chat. The assistant sets itself up and asks you for an API key (Project → AI agent → Keys on xpost):
Set up xpost so you can post to my social accounts. The REST API is https://xpost.to/api/v1 (spec at https://xpost.to/api/v1/openapi.json), authenticated with an
Authorization: Bearer YOUR_KEYheader. If you can run commands,npx xpostwraps it andnpx skills add xpost-to/xpost-agentinstalls the workflow. Ask me for my xpost API key and set it as XPOST_API_KEY before doing anything. Start with the accounts list. Creating a post sends it, unless my project holds posts for approval — never say a post is published until the receipt says so.
ChatGPT, Claude.ai, Claude Desktop
ChatGPT: xpost is in the plugin directory — find it, add it, sign in.
Claude with the xpost plugin installed: Customize → Connectors, find
xpost, press Connect, sign in — the plugin brings the connector, signed out.
Without the plugin: Customize → Connectors → Add → custom connector, paste
https://xpost.to/api/mcp, sign in.
No MCP client at all — the CLI
npx xpost login
npx xpost posts create -c "Shipped it." -a x:yourhandle
The xpost package on npm is the CLI, the local MCP server (npx xpost mcp)
and the skill's fallback in one. Docs: https://xpost.to/docs/cli
What the skill teaches
- Accounts first:
list_accountsnames every destination and what it can carry. get_posting_rulesis the option catalog per account; a key that is not there cannot be used.- Creating a post sends it; a project in copilot mode holds it (
pending_approval) instead. Either is a normal outcome — say which, print the link. - A guardrail refusal names the rule. Rewrite once. Never evade.
list_postsbefore assuming a post is still waiting — the person may have rejected it with a reason.- Receipts are per account: "delivered to X, failed on Instagram: reason".
Full text: skills/xpost/SKILL.md.
Docs
- Every MCP tool: https://xpost.to/docs/mcp
- REST API (OpenAPI): https://xpost.to/api/v1/openapi.json
- Set-up guides per client: https://xpost.to/docs
Self-hosting a client against your own instance
Replace https://xpost.to with your instance in the MCP address. For the
REST fallback set XPOST_URL and XPOST_API_KEY.
License
MIT. The skill and manifests here are free to copy and adapt; xpost itself is a hosted service.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.