Official remote MCP servers for integrating Facebook, Instagram, WhatsApp, and Meta Ads capabilities into your agentic workflows. ## Benefits - **One protocol, built for agent workflows.** Your agent connects to a Meta MCP server and discovers available tools at runtime. Tool descriptions, parameter schemas, and error responses are designed for LLMs to interpret correctly, which minimizes hallucination and failed tool calls. - **Standard authentication.** OAuth-based authentication with granular permission scopes per tool. Revoke access at any time from your Meta account settings. ## Available MCP servers The following table lists the available Meta MCP servers: | Server | Description | Type | |---|---|---| | [Meta Devtools MCP](/documentation/mcp/devtools-mcp) | Manage Meta apps, configure webhooks, check compliance, review app status, and search developer documentation. | Remote MCP | If you have feedback or want to see more tools, we want to [hear from you](https://forms.gle/6MUgLiWU2PWVtjeM6). ## Supported clients Below are clients validated by the Meta team for general availability. For configuration snippets per client, see the linked tools reference for each server. - Claude Desktop - Claude Code - ChatGPT (Web) - Codex App More clients coming soon. If your client supports remote HTTP MCP servers, it should work — even if not listed above. ## Get started The setup pattern is the same for every Meta MCP server. For per-client setup details, see the tools reference for each server. - **Add the server URL to your client.** Most clients have a Connectors or MCP settings panel. See the server's reference page for client-specific instructions. - **Sign in with your Meta account** when prompted. - **Select the apps and scopes** (read or manage) you want to grant on the consent screen. - **Verify.** Ask your agent what tools it has — you should see the server's tools listed. ## Security and best practices By accessing or using a Meta MCP server, you and the agent acting on your behalf are bound by the [Meta Platform Terms](https://developers.facebook.com/terms/). The Terms govern how data returned by Meta APIs may be stored, processed, shared, and displayed. You are responsible for ensuring that any agent or downstream system that consumes MCP responses complies with these obligations, including data retention limits, prohibited uses, and end-user transparency requirements. **Important:** Agents act on your behalf. Anything they can call with your granted scopes, they can call — including in response to malicious instructions hidden in tool outputs (prompt injection). Treat MCP servers like any third-party integration. ### Recommended practices - **Grant minimum scopes.** On the consent screen, only select scopes for tools you actually need. You can always add more later. - **Use separate apps for dev and prod.** Don't grant production app scopes to experimental agents. - **Audit periodically.** Review connected MCP servers at facebook.com > **Settings** > **Business Integrations** every few months. Revoke ones you no longer use. - **Watch for prompt injection.** Agents can be manipulated by content in webhooks, documents, or tool outputs. Don't grant write/manage scopes to agents that process untrusted input. ## More resources - [Meta Platform Terms](https://developers.facebook.com/terms) - [Responsible Platform Initiatives](https://developers.facebook.com/docs/resp-plat-initiatives) - [Business Integration](https://www.facebook.com/settings/?tab=business_tools) - [Model Context Protocol Spec](https://modelcontextprotocol.io) - [llms.txt](https://developers.facebook.com/llms.txt)