Continue

Continue is an open-source AI coding assistant that brings model-flexible AI help directly into a developer's editor. Distributed under the Apache 2.0 license as extensions for VS Code and JetBrains IDEs, its guiding principle has always been freedom of choice: developers bring their own models, from cloud providers like OpenAI, Anthropic, and Google to local models run through Ollama or LM Studio, so code can stay on their own machine.

Inside the editor, Continue offers a familiar set of capabilities. Autocomplete suggests inline code as you type, a sidebar Chat answers questions and helps debug without leaving the IDE, and Edit applies inline changes. Its Agent mode goes further, equipping the model with tools to read files, make changes across the codebase, and run commands so it can carry out multi-step tasks like large refactors or new features from a natural-language instruction, asking permission before each tool use by default. Custom agents, defined as instruction and rule files checked into a repository, let teams encode what good code means for their project, and the assistant can pull context from codebases, documentation, issues, and files.

In 2025 Continue expanded beyond the editor into what it called continuous AI: running AI review agents automatically on every pull request, with each review rule stored as a version-controlled Markdown file in the repository. Through that model, background agents plug into CI pipelines like GitHub Actions and run on schedules to triage issues and pull requests, address alerts from tools like Sentry, fix dependency vulnerabilities flagged by Snyk, and keep documentation current.

For organizations, Continue adds governance over which agents and model blocks developers can use, a proxy that lets members use API keys without seeing them, single sign-on, and an on-premises data plane that keeps code in the company's own environment. In 2026, Continue was acquired by Cursor, with its open-source codebase remaining available as a foundation others can build on. It has been a widely used choice for developers who wanted an open, model-agnostic alternative to closed coding assistants.