Scira MCP Chat is a Next.js-based chat application that integrates with Model Context Protocol (MCP) servers to expose external tools and capabilities to the chat assistant. The app streams AI responses and supports multiple providers via the AI SDK by Vercel, enabling you to switch between providers and MCP-enabled tools without leaving the chat interface. You can connect to MCP servers over HTTP or Server-Sent Events (SSE) transports, allowing tools such as search, code interpretation, workflows, and other provider-specific capabilities to be invoked from within the conversation. Use the settings icon next to the model selector to add MCP servers by supplying a name, choosing the transport type, and providing the server URL for HTTP or SSE as appropriate. Once added, you can enable a server for the current chat to expose its available tools to the model during that session.

Prerequisites:

• Node.js 16+ (or the version required by your environment)
• npm or pnpm/yarn for package installation

1. Clone the repository: git clone https://github.com/zaidmukaddam-scira/zaidmukaddam-scira-mcp-chat.git cd zaidmukaddam-scira-mcp-chat

2. Install dependencies: npm install

   or use yarn/pnpm if preferred

3. Configure environment (optional):

   • If the app requires any API keys or MCP endpoints, configure them in a .env.local file as needed by the project (the README does not specify required env vars).

4. Run the development server: npm run dev

   or: yarn dev / pnpm dev

5. Open the app in your browser: http://localhost:3000

Notes:

• The MCP servers are connected via the chat UI and do not require a separate local server in most setups. If you deploy a local MCP endpoint, ensure it is reachable from the app and that you provide the correct HTTP/SSE URL via the UI.

Tips and common considerations:

• MCP servers can expose tools like search, code interpretation, and automation; you can mix multiple MCP providers (e.g., Composio, Zapier, Hugging Face) by adding them as separate servers in the chat settings.
• Use the HTTP or SSE transport options to suit the latency and tooling requirements of the MCP provider.
• Streaming responses enable progressive rendering of results; ensure your UI handles partial chunks if using streaming.
• If you encounter connection issues, verify that the MCP server URL is reachable from your deployment environment and that any required authentication headers are configured in the server settings.
• Environment variables can be used to toggle features or set log levels for debugging (e.g., MCP_ENABLED, MCP_LOG_LEVEL).
• This app leverages the AI SDK by Vercel for provider-agnostic AI orchestration; check provider-specific limits and rate quotas in your MCP subscriptions.