Algolia MCP

Algolia's MCP server exposes the search platform's APIs to AI assistants, so an agent can work with indices the same way an engineer would through the dashboard or a client library. It can run queries against an index and return ranked hits, inspect the records and attributes behind those results, list the applications and indices available on an account, and read or change configuration such as searchable attributes, custom ranking, synonyms, rules and Recommend models. The practical value is debugging and tuning relevance in a conversation. Instead of manually reproducing a query, diffing settings and reading analytics dashboards, a developer can ask why a record ranks where it does, have the assistant compare index configurations, or pull query analytics to spot searches that return nothing. Ingestion and index management operations let the same session push or update data while a search experience is still being built. It targets developers and search engineers building Algolia-powered discovery in web and mobile products, plus the teams who maintain relevance after launch. The server runs as a local or remote MCP endpoint authenticated with Algolia application credentials, so what an agent can reach is bounded by the API key scopes you hand it. It works with any MCP client, which in practice means a coding agent sitting next to the code that calls Algolia.