Upstash MCP

Upstash MCP is the Model Context Protocol server for Upstash, the serverless data platform behind managed Redis, message queueing, and vector storage. It turns account administration and data access into tools an AI assistant can call directly, so an agent working inside an editor or a chat client can inspect infrastructure and act on it without the operator switching over to the Upstash console. The server exposes the database lifecycle: listing databases, creating and deleting them, reading configuration such as region and plan settings, and pulling usage counters like command throughput, bandwidth, and storage over a chosen time window. It also exposes data-plane access, letting an agent run Redis commands against a selected database to read keys, inspect data shapes, and debug what an application actually wrote. Queue and scheduling resources are reachable the same way, so message topics, schedules, and delivery history can be reviewed alongside the database that feeds them. It fits teams already running Upstash in production who want an assistant that answers operational questions with real data instead of guesses: which database is absorbing the traffic, whether a cache key exists, why a scheduled job never fired. Because the same credentials that drive the console drive the server, scoping the API token is the main control on what a connected agent can reach.