
Continue
Open-source, model-agnostic AI code assistant for VS Code and JetBrains that you configure to your own stack
Gallery
About Continue
Continue is an open source AI code assistant that operates directly inside VS Code and JetBrains IDEs. Rather than forcing you into a separate application or browser window when you need AI assistance with code, Continue integrates that capability into the editor where you already spend your working hours. The project was built on the philosophy of amplifying developers rather than attempting to replace them, treating artificial intelligence as a collaborative partner that handles tedious tasks while humans retain control over architecture, design decisions, and the higher order thinking that still requires human judgment. The entire codebase remains freely available for inspection, modification, and extension.
The assistant operates through several integrated interaction modes designed for different types of coding tasks. Chat provides a conversational interface for discussing code, asking questions about how something works, requesting explanations of complex sections, or brainstorming approaches to problems you're trying to solve. You can highlight a block of code and ask the AI to explain it, suggest improvements, identify potential bugs, or propose alternative implementations. The conversation maintains context across exchanges, so you can iterate on ideas without constantly re establishing what you're working on.
Edit mode handles in place code modifications without the friction of copying content between windows. You describe what you want changed in natural language, and Continue applies the edits directly in your active file. This streamlines the common pattern of asking for help and then manually implementing suggestions. The AI proposes changes, you review them in your editor's native diff view, and you accept or reject modifications with full visibility into exactly what will change. For refactoring tasks, updating function signatures, or applying consistent changes across sections of code, edit mode turns verbal instructions into immediate implementation.
Autocomplete provides real time suggestions as you write, anticipating what comes next and offering completions for boilerplate, function bodies, and repetitive patterns. The suggestions adapt to your codebase context, meaning they reflect the conventions, naming patterns, and structures already present in your project rather than generic templates that might not fit. This acceleration applies throughout your typing without requiring explicit invocation.
Agent mode tackles multi step coding tasks through ongoing collaboration rather than single shot responses. You can describe a larger goal, and the agent works through the problem with you, breaking it into manageable pieces and executing incrementally. This proves valuable for refactoring sessions that touch multiple files, implementing features that require coordinated changes across a codebase, or exploring unfamiliar code where you need guidance on where to make changes and what the downstream implications might be.
For developers who prefer terminal based workflows, the command line tool called cn extends AI coding assistance outside the graphical IDE. If you're working in a terminal environment or automating development tasks through scripts, cn brings the same underlying capabilities to that context.
Continue supports multiple AI model providers through a flexible configuration system. You can connect to OpenAI, Claude, Ollama for local inference, and various other providers depending on your requirements. The configuration allows different models to power different functions independently. You might use a fast model for autocomplete where speed matters most, while directing complex conversations to a more capable model. Local model support means codebases with confidentiality requirements can stay entirely on your machine without transmitting anything to external services.
The context system enables sophisticated awareness of your project beyond the immediate file. Model Context Protocol integration, custom rules, and documentation connections let you configure how Continue understands your codebase structure, your team's conventions, and relevant external information. Custom prompts shape the AI's responses to match established patterns or personal preferences. This configurability distinguishes Continue from more rigid tools that impose predetermined workflows.
Continue recently joined Cursor, the AI native code editor that has attracted significant developer attention. The open source codebase continues to exist as community infrastructure, but the core team now develops under the Cursor organization. For developers who want extension based AI assistance without adopting an entirely new editor, Continue remains compelling. You keep your existing VS Code or JetBrains setup with all your keybindings, themes, and extensions intact while gaining AI capabilities layered on top.
Key Features
- Open-source core under the Apache 2.0 license
- VS Code and JetBrains extensions plus a Continue CLI
- Model-agnostic: bring your own keys for OpenAI, Anthropic, local Ollama, and custom endpoints
- Agent, Chat, Edit, and Autocomplete modes inside your editor
- Continue Hub for building and sharing custom assistants, rules, and prompts
- Source-controlled AI checks you can enforce in CI
Pros & Cons
What we like
- No vendor lock-in, you pick and swap the underlying model
- Can run fully local through Ollama for privacy-sensitive code
- Deeply configurable assistants, rules, and context providers
- Free, self-hostable open-source core
Room for improvement
- Real setup and config effort before it feels tuned to your workflow
- You supply and pay for your own model API keys (BYO key)
- The JetBrains plugin is now community-maintained, with the CLI as the recommended path
- Hub team features (private rules, SSO, BYOK) sit behind paid plans
Frequently Asked Questions
What is Continue?
Is Continue free?
Who is Continue for?
How is Continue different from GitHub Copilot?
Best For
Featured in
Alternatives to Continue

Cline
Open-source autonomous coding agent for VS Code that runs on your own API key with no subscription
Aider
Open-source AI pair programmer that lives in your terminal, edits your code, and auto-commits to git
GitHub Copilot
AI pair programmer from GitHub with inline completions, chat, agent mode, and model choice across IDEs
Cursor
VS Code fork built around an AI pair-programmer
Reviews (9)
Quietly excellent
Almost a year on Continue now, no plans to leave. Their take on deeply configurable assistants, rules, and context providers is genuinely good.
Worth the price of admission
Picked Continue for the price, stayed for the quality. Their take on can run fully local through ollama for privacy-sensitive code is genuinely good. No regrets so far.
Worth the price of admission
Almost a year on Continue now, no plans to leave. Real selling point for me was open-source core under the apache 2.0 license. Genuine strength is that it gets out of the way and lets me work. It fits well for enforcing source-controlled ai code checks in ci. No regrets so far.
Genuinely impressed
Tried Continue on a side project first, then rolled it out everywhere. Real selling point for me was free, self-hostable open-source core. Found it works best for pairing a frontier or local model with your editor without provider lock-in. Recommending it to people in a similar spot.
It just works
Onboarded the whole team to Continue in an afternoon. Nothing flashy, it simply pulls its weight every day. The defaults are sensible, so I was not fighting settings on day one. It earns its place in my stack.
The kind of tool you forget you are paying for
Tried Continue on a side project first, then rolled it out everywhere. It slotted into my routine without much fuss. Found it works best for standardizing shared assistants, rules, and prompts across a team.
It just works
Picked Continue for the price, stayed for the quality. Their take on agent, chat, edit, and autocomplete modes inside your editor is genuinely good. Found it works best for running a private, local-only coding assistant via ollama. Worth the price for what I get out of it.
Worth the price of admission
Three months of Continue later, here is what holds up. Real selling point for me was source-controlled ai checks you can enforce in ci. Performance has been steady even when I lean on it hard. No regrets so far.
Good, with a few caveats
Hadn't planned on switching, but Continue was hard to ignore. Real selling point for me was continue hub for building and sharing custom assistants, rules, and prompts. It handles the boring parts so I can focus on the work that matters. It fits well for running a private, local-only coding assistant via ollama. It would be a five if not for hub team features (private rules, sso, byok) sit behind paid plans. Glad I made the switch.
Badge builder
Add Continue to your website
Choose a badge style and size, preview it here, then copy the generated HTML. Badge images are self-contained SVGs and do not require an external script.
<a href="https://toolindex.net/tools/continue-dev?ref=badge" target="_blank" rel="noopener">
<img src="https://toolindex.net/badge/continue-dev/medium.svg" alt="Continue - Listed on Tool Index" width="180" height="50" />
</a> How to use the badge
- 1. Pick the style, size, and theme that fit your layout.
- 2. Copy the generated HTML from the code block.
- 3. Paste it into your footer, homepage, or press page.
Standard badge available
The standard listing badge is available now. Score and circle badges are limited to tools currently ranked in the top 10 of a category.
Badge clicks return visitors to this profile with a referral tag so the source remains identifiable.
Related Tools

Bolt.new
Prompt-to-deployed full-stack app inside the browser

Tabnine
Privacy-first enterprise AI code assistant you control, with SaaS, VPC, on-prem, or fully air-gapped deployment
Cursor
VS Code fork built around an AI pair-programmer

Skeezo
Agent-native macOS IDE for coordinating Claude Code and AI agents
Work on Continue? Request listing access or correction