This MCP repository aggregates several Cardea services, each implemented as an MCP server. The included servers expose distinct capabilities: a Cardea Calculator for computational tools, a Weather service for weather-related queries, a Qdrant-based vector data service, a keyword search utility, an Elastic-backed search/indexing server, a TiDB-backed data store, a GitHub-related data service, and a Web Search interface. Each server follows the MCP protocol, allowing you to send requests and receive structured responses through the same interface. To use, deploy the MCP servers in your environment and connect to them via your MCP client, routing requests to the appropriate server by name. Consult the individual subproject READMEs for endpoint details and supported MCP methods per service (e.g., calculation, search, indexing, or data retrieval).

Prerequisites:

• A compatible runtime environment (Node.js for the Node-based servers, or the language/runtime used by each subproject as documented in their READMEs).
• Access to the repository containing the Cardea MCP Server suite.
• Basic familiarity with MCP protocol and your environment’s networking rules.

Installation steps (per server):

1. Clone the repository containing the MCP server suite:

   git clone https://github.com/cardea-mcp/cardea-mcp-servers.git cd cardea-mcp-servers

2. Install dependencies for each submodule. Examples (adjust per server language):

   • If a server uses Node.js: cd cardea-calculator && npm install cd ../cardea-weather && npm install

     repeat for other Node-based servers

   • If a server uses Python/uvicorn or another runtime, follow that subproject’s README for environment setup and installation instructions.

3. Configure runtime specifics if needed (see env vars below).

4. Start the servers. Examples (Node-based): node path/to/server.js

   Or use a repository-provided start script if available

5. Verify the servers are reachable via your MCP client and begin sending requests.

Note: Some servers may have additional prerequisites or require separate setup (e.g., a database instance, a vector store, or external services). Always refer to the individual server READMEs for exact steps.

Environment variables and configuration options are typically documented per server in their READMEs. Common considerations include: - NETWORK_HOST and NETWORK_PORT for exposing MCP endpoints.

• AUTH or API_KEY for access control if the server supports it.
• Paths to model weights, databases, or vector stores if required by the server (e.g., Redis, TiDB, Elastic, Qdrant).
• Timeouts and resource limits to prevent long-running requests from blocking the MCP bus. If you encounter issues, check the server logs for startup errors, verify MCP client compatibility, and ensure that the target endpoints are reachable from your client environment.