MCP Server

Connect Claude, Cursor, ChatGPT, and other AI agents to PostPeer via the Model Context Protocol.

PostPeer runs a hosted MCP (Model Context Protocol) server at https://mcp.postpeer.dev/mcp. Add one URL to your AI client and it gets 34 tools covering every PostPeer endpoint — publishing, scheduling, analytics, integrations, profiles, AI helpers, and more.

Nothing to install. Nothing running locally.

What's MCP?

Model Context Protocol is an open standard from Anthropic for letting AI assistants call external tools. When you add PostPeer's MCP server to a compatible client, the AI can publish posts, schedule content, fetch analytics, and manage your social accounts directly.

Connect your AI client

You'll need a PostPeer access key — grab one from the Dashboard.

Claude Desktop / Claude Code / Cursor

Add the following to your MCP config:

claude_desktop_config.json / .cursor/mcp.json

{
	"mcpServers": {
		"postpeer": {
			"url": "https://mcp.postpeer.dev/mcp",
			"headers": {
				"Authorization": "Bearer YOUR_POSTPEER_API_KEY"
			}
		}
	}
}

Restart the client. PostPeer's tools (create_post, list_integrations, get_analytics, etc.) will appear in the available-tools list.

ChatGPT (and other Streamable-HTTP MCP clients)

Same shape — point the client at https://mcp.postpeer.dev/mcp and pass Authorization: Bearer YOUR_POSTPEER_API_KEY. If your client doesn't support custom headers, append ?access_key=YOUR_POSTPEER_API_KEY to the URL instead.

Available tools

The MCP server exposes one tool per public API operation. Common ones:

Tool	What it does
`list_integrations`	Show connected social accounts
`create_post`	Publish or schedule a post to one or more platforms
`list_posts` / `get_post` / `delete_post`	Browse and manage posts
`list_scheduled_posts` / `reschedule_post` / `cancel_scheduled_post`	Manage the publish queue
`get_analytics`	Cross-platform post metrics
`ai_write_content` / `ai_generate_image`	AI helpers for captions and images
`get_pinterest_boards`	Required before posting to Pinterest
`get_tiktok_creator_info`	Required before posting to TikTok
`create_profile` / `list_profiles` / `update_profile`	Multi-tenant profile management
`create_app` / `list_apps`	Manage BYOK OAuth apps

A full list lives on the MCP repo: github.com/geiger01/postpeer-mcp.

Try it

Once connected, ask your AI client things like:

"List my connected social accounts."
"Draft a tweet announcing our new product and schedule it for tomorrow at 9am Eastern."
"How did my last 5 LinkedIn posts perform?"
"Generate an image for a post about Q3 results and publish it to Twitter and LinkedIn."

The AI calls the relevant tools, PostPeer publishes, and you get a confirmation back inline.

Platform prerequisites

Before publishing to certain platforms, the AI should call a helper tool first:

Pinterest — call get_pinterest_boards to pick a board, then pass the board id as platformSpecificData.boardId on create_post. Pins must land on a board.
TikTok — call get_tiktok_creator_info and pass platformSpecificData.privacyLevel using one of the returned privacyLevelOptions. Required by TikTok's content-sharing rules.

create_post's tool description points the model at these prereqs automatically, so you shouldn't need to prompt for them.

Authentication

The MCP server accepts your access key three ways (priority order):

Authorization: Bearer <key> — recommended.
x-access-key: <key> — same header the REST API uses.
?access_key=<key> query string — for clients without custom-header support.

The MCP server never validates the key itself — it forwards every call to api.postpeer.dev, which handles auth, credits, and rate limiting. So credit usage, plan limits, and access-key permissions behave identically to REST.

Self-hosting

If you need the MCP on your own infrastructure (compliance, custom domain, etc.), the source is open: github.com/geiger01/postpeer-mcp. It's a small Node + Express + @modelcontextprotocol/sdk server — runs anywhere Node runs (Railway, Fly, Render, your own VPS).

Troubleshooting

401 Unauthorized — your access key is missing or invalid. Confirm the Bearer token is set in the client config and that the key works against the REST API.
429 Too Many Requests — your plan's rate limit. Either upgrade or space the calls out.
Tools don't appear in the client — restart the AI client after editing the config. Some clients also need MCP enabled explicitly in settings.

On this page