Skip to main content
The Postsiva Unified MCP lets AI agents call Postsiva tools instead of hand-writing REST requests. Each tool runs the same backend logic as the website/WhatsApp agent.

​
How it works

  1. Your MCP client connects to the right endpoint:
    • CLI (Cursor / Claude Code): https://mcp.postsiva.com/mcp
    • Web (Claude.ai / ChatGPT): https://mcp.postsiva.com/web/mcp
  2. The agent calls tools (publish, get_user_posts, upload_media, …).
  3. Each call resolves workspace_id from the key/OAuth and invokes the shared tool layer.
  4. Results are JSON returned to the agent.
There is no separate MCP business logic — quotas, plan gates, and platform rules match the REST API.

​
Endpoints

  • Transport: Streamable HTTP (stateless)

​
Who it’s for

  • Claude.ai / ChatGPT — web MCP with in-chat upload widget
  • Cursor / Claude Code — CLI MCP with curl upload
  • n8n / automation — AI Agent node with MCP
  • Custom agents — any MCP-compatible client

​
Requirements

  • Pro plan (mcp_enabled)
  • Valid workspace API key or OAuth connector
  • Social accounts connected in that workspace

​
Tool count

Core tools on both servers (accounts, posting, queue, posts, analytics, comments, AI, platform connect). Upload differs (iframe vs curl). No MCP web-search tool (hosts provide search). See the full Tool catalog.

​
vs REST API

Use REST when you know the exact call; use MCP when an agent should decide which action to take.

​
Next steps

Connect & setup

Cursor, Claude, headers

Tool catalog

All tools

Cursor guide

Postsiva MCP in Cursor

n8n

AI Agent + MCP