Continue
The open-source AI code assistant for VS Code and JetBrains.
What it does
Why it matters
Continue proved that AI code assistance doesn't have to be a walled garden. Open source, working with any model provider (including local ones), and embedded in VS Code and JetBrains — it's the choice for developers and teams who want to see how the sausage is made, or who can't send code to a third party.
Engineering notes
- Model-agnostic by design. Configure Claude, GPT, Ollama, or anything OpenAI-compatible — switch models without switching tools.
- Open source, genuinely. The codebase is inspectable and forkable, which matters for security reviews and custom deployments.
- Context providers. Explicitly attach files, docs, URLs, and terminal output to your prompts — a thoughtful answer to the context problem.
Limitations
- Setup takes more work than commercial alternatives — you configure models yourself.
- Feature velocity trails the venture-funded competitors.
- Quality depends heavily on which model you pair it with.
Related projects
Bun
An all-in-one JavaScript runtime, bundler, test runner, and package manager.
Bun collapsed the Node.js toolchain — runtime, transpiler, bundler, test runner, package manager — into a single Zig binary that's dramatically faster at nearly everything. Cohesion as a performance strategy.
Cursor
The AI-first code editor, forked from VS Code.
Cursor took VS Code's familiar surface and rebuilt the AI integration underneath: tab completion that predicts your next edit, a composer for multi-file changes, and agent mode that runs terminal commands.
GitHub Copilot
The AI pair programmer that started it all.
Copilot pioneered AI-assisted coding at scale: inline completions trained on public code, now expanded into chat, agent mode, and code review. The baseline every other tool gets measured against.
Canonical URL: https://notvibescoded.com/projects/continue-dev