Skip to content
| Marketplace
Sign in
Visual Studio Code>Other>PilotNew to Visual Studio Code? Get it now.
Pilot

Pilot

nartaliti

|
69 installs
| (0) | Free
Host your own coding assistant instantly
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Pilot

Pilot is your self hosted, privacy first coding assistant for Visual Studio Code. It brings AI powered chat and inline code completions right into your editor helping you code, debug, and explore your project faster. Best of all, your code stays yours no data ever leaves your machine.

Features

  • ✨ Inline Code Completions Get real time suggestions based on your code context, powered by your own models.

  • 💬 Chat with AI Ask questions, get explanations, or debug problems using a smart developer assistant.

  • 📁 Workspace Indexing Boost accuracy by indexing your project so the assistant understands your codebase better.

  • 🧠 Customizable System Prompts Tune how Pilot behaves, both in completions and chat.

  • ⌨️ Keyboard Shortcuts

    • Ctrl+Shift+Space / Cmd+Shift+Space: Rerun inline completion

    • Ctrl+Alt+P / Cmd+Alt+P: Open chat

    • Ctrl+Alt+I / Cmd+Alt+I: Index workspace


Requirements

To use Pilot, you must run a local or remote Ollama server with 2 types of models, a normal LLM for coding and an Embedding LLM for indexing.

The ones I recommend are:

  • codellama:instruct (for inline completion and chatting)

  • nomic-embed-text:latest (for embedding workspace context)

You can also use other models.

Install them with:

ollama pull codellama:instruct
ollama pull nomic-embed-text:instruct

Then, set the server URL and model names in your VS Code settings.

⚠️ Heads up: If your machine isn’t powerful enough to run these models smoothly, you may notice delays during chat, completion, or indexing


Extension Settings

You can configure Pilot through your VS Code settings (Cmd/Ctrl + ,, then search for "Pilot"):

  • pilot.serverUrl: URL of your local/remote Ollama server

  • pilot.model: Model used for inline code suggestions

  • pilot.embeddingModel: Model used to understand your project context

  • pilot.autoCompleteSystemMessage: Customize how inline completions behave

  • pilot.chatSystemMessage: Adjust the tone/personality of the chat assistant


Known Issues

  • 🚫 NONE

Release Notes

1.0.0

  • Initial release of Pilot

  • Inline code completions

  • AI powered chat

  • Workspace indexing

  • Customizable prompts


Contributing

Excited to have contributors on board! While contribution guidelines are still in progress, feel free to try out the extension and report any issues.


License

MIT © Nart Aliti

Pilot is all about giving you control over your AI assistant—right inside your favorite editor. Happy coding! 🧑‍💻

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2025 Microsoft