Continue.dev's core idea is configurability: instead of a fixed assistant tied to one model provider, it lets developers define their own setup through reusable building blocks, models, prompts, rules, MCP tool connections, and custom context sources, in a single config file. That makes it possible to build an assistant tuned to a specific team's stack rather than accepting whatever a vendor decided to bundle.
It offers four modes, Chat, Autocomplete, Edit, and Agent, covering everything from inline suggestions to full multi-step refactoring, and it works with essentially any LLM: OpenAI, Anthropic, local models through Ollama, Mistral, and others. Enterprises get on-premises or cloud deployment with centralized governance over model selection, tool permissions, and usage tracking.
Licensed under Apache 2.0 and free to use, it has built a genuinely active open-source community, with around 27,800 GitHub stars and an engaged Discord, driven mostly by developers who want to own their AI assistant's configuration rather than work within a closed product's defaults.