xpoz

Xpoz is a remote MCP server that lets you search across Twitter, Instagram, Reddit, and TikTok through natural language queries, all authenticated via Google OAuth. The server exposes a streamable HTTP endpoint, so you don’t need to install anything locally to access its capabilities. Once configured in your MCP client, you can issue queries to search for posts, profiles, and topics across the supported platforms, generate CSV exports up to 500k rows, and paginate large result sets. The available tools cover a broad set of actions (for example, on Twitter you can search users, fetch profiles, search posts by keywords, and retrieve comments or engagement data; on Instagram and Reddit, you can perform analogous searches and retrieve user/profile information; TikTok is listed as coming soon). Use natural language prompts to direct what data you want and apply field selections to limit the response to the data you actually need. The service uses OAuth 2.1 for authentication, so you’ll authenticate via Google OAuth when prompted by your client.

To use Xpoz in your MCP-enabled agent, add the remote endpoint as shown in the Quick Start: configure the MCP client to point to the streamable HTTP URL, authenticate as needed, and then issue queries like “Show me recent posts about AI ethics from Twitter and Reddit” or “Export top influencers mentioning climate change” and the server will handle the cross-platform retrieval and export tasks. If you’re using OpenClaw or other tooling, you can install the xpoz-social-search integration (as shown in the README) and follow the prompts to authorize access.

Note: Xpoz is a hosted, remote MCP server. No local installation is required. You only need a client that supports MCP and a network path to the remote endpoint.

Prerequisites

• An MCP client or orchestrator that can consume streamable HTTP MCP servers
• Internet access from your client to the remote endpoint
• OAuth credentials (Google OAuth flow will be triggered by the client during first use)

Steps

1. Prepare the MCP configuration for your agent. Use the remote streamable HTTP endpoint provided by Xpoz: { "mcpServers": { "xpoz": { "type": "streamable-http", "url": "https://mcp.xpoz.ai/mcp" } } }
2. Save this configuration in your MCP client’s config file or UI, following your tool’s instructions for adding a new MCP server.
3. If your tool requires, trigger the Google OAuth flow when prompted to authenticate access to the platforms via Xpoz.
4. Test a basic query such as: "What are the latest posts mentioning climate change from Twitter and Reddit?" or use any supported natural language prompt.
5. (Optional) If your workflow uses OpenClaw or ClawHub, install the xpoz-social-search package as shown in the README and authenticate when prompted.

If you’re integrating into an automated workflow, you can also script a simple query against the streamable HTTP endpoint once authenticated by your MCP client.

Tips and considerations:

• The service supports server-side pagination and async exports; use status polling endpoints to monitor long-running exports.
• CSV exports can handle up to 500K rows in a single download; if you need more, consider chunked exports or iterative queries.
• Apply field selection to reduce payload size and improve performance for large queries.
• The OAuth flow is Google-based; ensure your environment can complete the browser-based or headless OAuth as your client supports.
• If you encounter issues with platform availability (e.g., TikTok is marked as coming soon), check the service status or documentation for updates on supported tools.
• No API keys are required to use the remote MCP server itself; authentication is handled via OAuth via your MCP client.