crosspost-mcp

MCP server that lets AI agents cross-post content to Telegram, Discord, Mastodon, Bluesky, and Reddit — using free APIs only.

Why crosspost-mcp

AI agents can now draft, research, and refine content end-to-end — but publishing to social media is still the manual bottleneck. crosspost-mcp closes that gap by exposing posting tools over the Model Context Protocol, so Claude Desktop, Cursor, and other MCP-compatible clients can publish on your behalf with a single prompt.

Related MCP server: Social CLI MCP

Demo

Claude calls post_to_discord and the message appears in your channel instantly:

Features

• Free APIs only — no paid SaaS middlemen; Telegram Bot API, Discord webhooks, and upcoming platform APIs are all free-tier friendly

• Async parallel posting — post_to_all fans out to every configured platform concurrently

• Type-hinted — full Python type annotations on tools and platform modules

• Dataclass configs — frozen dataclasses per platform with clear missing-variable errors

• Easy to extend — add a new platform by dropping a module in platforms/ and registering a tool

Supported Platforms

🚀 Quick Start

Clone & install

git clone https://github.com/Aleksey-Panf/crosspost-mcp.git
cd crosspost-mcp
pip install -e .

Setup credentials

Copy the example env file and fill in your tokens:

cp .env.example .env

Edit .env with your Telegram bot token, Discord webhook URL, and any other platform credentials you plan to use.

Add to Claude Desktop

Open your Claude Desktop MCP config:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

• Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the server entry (adjust the path to your clone):

{
  "mcpServers": {
    "crosspost-mcp": {
      "command": "crosspost-mcp",
      "cwd": "/path/to/crosspost-mcp",
      "env": {
        "TELEGRAM_BOT_TOKEN": "your_bot_token",
        "TELEGRAM_CHAT_ID": "your_chat_id",
        "DISCORD_WEBHOOK_URL": "https://discord.com/api/webhooks/..."
      }
    }
  }
}

Tip: If crosspost-mcp is not on your PATH, use the full path to the executable inside your virtualenv, or point command to python with args: ["-m", "crosspost_mcp.server"].

Cursor users can add the same block under Settings → MCP → Add server.

Restart Claude Desktop

Fully quit and relaunch Claude Desktop (or reload MCP in Cursor). The post_to_telegram, post_to_discord, and post_to_all tools should appear in the tool list.

Usage Examples

Give your AI agent natural-language instructions — it picks the right MCP tool automatically.

Example 1 — Discord only

Post to Discord: "v0.1.0 is live — crosspost-mcp now supports Telegram and Discord webhooks."

Claude calls post_to_discord with the message text and returns the Discord message_id and channel_id.

Example 2 — Telegram only

Announce on Telegram that we shipped parallel cross-posting. Use bold for the version number.

Claude calls post_to_telegram with HTML formatting (<b>v0.1.0</b>) and confirms the message_id.

Example 3 — Cross-post everywhere

Cross-post this to all platforms: "New blog post is up — link in bio."

Claude calls post_to_all, which posts to Telegram and Discord in parallel and returns a combined status (ok, partial, or error) with per-platform results.

🔧 Tool Reference

post_to_telegram

Post a message to the configured Telegram channel.

Example response:

{
  "platform": "telegram",
  "status": "ok",
  "message_id": 42,
  "chat_id": -1001234567890
}

post_to_discord

Post a message to the configured Discord channel via webhook.

Example response:

{
  "platform": "discord",
  "status": "ok",
  "message_id": "1234567890123456789",
  "channel_id": "9876543210987654321"
}

post_to_all

Cross-post the same message to all configured platforms in parallel.

Individual platform failures do not abort the others. Each failure is reported with status: "error" and an error message.

Example response:

{
  "status": "ok",
  "results": [
    {
      "platform": "telegram",
      "status": "ok",
      "message_id": 42,
      "chat_id": -1001234567890
    },
    {
      "platform": "discord",
      "status": "ok",
      "message_id": "1234567890123456789",
      "channel_id": "9876543210987654321"
    }
  ]
}

Getting Credentials

Telegram

1. Open Telegram and message @BotFather.

2. Send /newbot, follow the prompts, and copy the bot token.

3. Add the bot as an admin to your target channel with Post Messages permission.

4. Get the chat ID:

   • For public channels, use @channelname (include the @).

   • For private channels, forward a channel message to @userinfobot or call getUpdates on the Bot API after posting in the channel.

5. Set TELEGRAM_BOT_TOKEN and TELEGRAM_CHAT_ID in your .env.

Discord

1. Open your Discord server and go to the target channel.

2. Click the gear icon → Integrations → Webhooks → New Webhook.

3. Name the webhook, select the channel, and click Copy Webhook URL.

4. Set DISCORD_WEBHOOK_URL in your .env.

No bot application or OAuth flow required — webhooks are the simplest path for AI agent posting.

Project Structure

crosspost-mcp/
├── assets/                          # Screenshots and demo images
├── src/
│   └── crosspost_mcp/
│       ├── __init__.py
│       ├── config.py                # Dataclass configs loaded from .env
│       ├── server.py                # FastMCP server and tool definitions
│       └── platforms/
│           ├── __init__.py
│           ├── telegram.py          # Telegram Bot API client
│           └── discord.py           # Discord webhook client
├── tests/
├── .env.example                     # Credential template
├── pyproject.toml
└── README.md

Roadmap

• Telegram support

• Discord support

• Parallel cross-posting

• Mastodon support (v0.2.0)

• Bluesky support (v0.3.0)

• Reddit support (v0.4.0)

• Media/image attachments

• Scheduled posting

Contributing

Contributions are welcome — bug reports, platform modules, and documentation improvements all help. Open an issue to discuss larger changes before submitting a pull request.

Custom MCP Development

I build production-ready MCP servers on demand for custom platforms and workflows.

• Delivery: 3–7 days

• Workflow: Async-only, no calls

• Payment: USDT

• Contact: Telegram @AlexAi14 or open a GitHub issue

License

MIT — see LICENSE for details.

Topics: mcp · model-context-protocol · fastmcp · telegram-bot · discord-webhook · ai-agents · claude · llm-tools