Octopus MCP Server is an AI-first gateway that enables AI assistants to access and interact with your Octopus Deploy instance through the MCP (Model Context Protocol). It exposes a curated set of tools that let an assistant inspect, query, and diagnose Octopus resources, transforming it into a powerful DevOps companion. Tools are organized into toolsets (core, projects, deployments, releases, tasks, tenants, kubernetes, machines, certificates, accounts) and can be enabled or restricted via the --toolsets flag. By default, the server runs in read-only mode for security, and logs can be directed to console or a file. It supports installation via Docker or Node (npx), with configuration through environment variables (OCTOPUS_API_KEY, OCTOPUS_SERVER_URL) or CLI arguments. Most tools leverage stable Octopus Server APIs dating back to 2021.1, with newer tools indicating minimum supported versions. The MCP server is in Early Access and subject to breaking changes. It comes with security guidance about sensitive data in logs and the risk of prompt-injection, and suggests least-privilege accounts. See the documentation for supported use-cases, prompts, and compatibility notes.