What is MCP?
MCP is an experimental Model Context Protocol server that enables AI assistants to interact with Algolia APIs for search, indexing, analytics, and administration tasks. It is built for developers and AI enthusiasts who want to control Algolia directly from tools like Claude Desktop or Ollama.
How to use MCP?
Clone the repository, build the Go binary, then configure it as an MCP server in Claude Desktop by adding the binary path and Algolia credentials to the mcpServers JSON settings. Optionally set MCP_ENABLED_TOOLS to restrict which tools are available, or switch to SSE transport via MCP_SERVER_TYPE. You can also run MCP with the MCP Inspector for debugging or with Ollama using the mcphost CLI.
Key features of MCP
- Enables natural‑language interaction with Algolia APIs via MCP.
- Supports both stdio and SSE transport protocols.
- Includes tools for search, indexing, analytics, A/B testing, and recommendations.
- Allows selective tool activation through the
MCP_ENABLED_TOOLS environment variable.
- Works with Claude Desktop, Ollama, and any MCP‑compatible client.
Use cases of MCP
- Search products in an Algolia index with filters like brand and price.
- Add, update, or delete records in an index via AI prompts.
- Modify index settings, such as searchable attributes or ranking rules.
- Run analytics or A/B testing experiments without a dedicated UI.
- Use local AI models (via Ollama) to manage Algolia data.
コメント