Skip to content
| Marketplace
Sign in
Visual Studio Code>Programming Languages>ReticleNew to Visual Studio Code? Get it now.
Reticle

Reticle

Roboalchemist

|
26 installs
| (0) | Free
Local-first BYOK FIM autocomplete with multi-line ghost text for OpenAI-compatible completion endpoints.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Reticle logo: code brackets and a completion sparkle

Reticle + Reticle MLX

Fast, private code completion on your own endpoint.
A native Apple Silicon model app and a local-first VS Code extension, designed to work together.

VS Code Marketplace version Open VSX version Build status MIT license

Get started · Mac app + extension · Extension only · Provider guides · Documentation


Reticle offers completely local, Cursor-style tab autocomplete in VS Code.

The project has two primary parts:

  • Reticle is the VS Code extension that connects your editor to Reticle MLX or another compatible POST /v1/completions endpoint.
  • Reticle MLX is the native menu-bar app for running, switching, benchmarking, and monitoring completion models on an Apple Silicon Mac.

With Reticle MLX, model traffic stays entirely local on your Mac. There is no hosted Reticle account or telemetry. Aside from model downloads you initiate, Reticle MLX's only internet use is Sparkle's minimal update check.

Choose your setup

If you… Start here
Use an Apple Silicon Mac and want the easiest complete setup Install Reticle MLX, then let the app install and verify the VS Code extension.
Already run a compatible local or hosted completion endpoint Install the Reticle VS Code extension and connect it to that endpoint.

Easiest setup: Reticle MLX + VS Code

Reticle MLX is the recommended path for Apple Silicon MacBooks and desktops. It handles the model runtime, downloads, service lifecycle, logs, health checks, benchmarking, and VS Code setup without requiring a terminal after installation.

  1. Install and open the signed, notarized Mac app:

    brew install --cask roboalchemist/tap/reticle-mlx
    open -g -a "Reticle MLX"
    

    You can also download the DMG or ZIP from the latest GitHub release.

  2. Open Settings… → Models, download a model, and click Select. The recommended model appears first; lighter, faster, and larger alternatives are listed below it.

  3. Open VS Code Setup, click Install VS Code Extension, then run VS Code Doctor. The doctor verifies the installed extension, endpoint, selected model, FIM format, and a real suffix-dependent completion.

  4. Open a code file in VS Code and start typing. Reticle suggests automatically; press Tab to accept. Use Option+\ to request a suggestion immediately.

The menu-bar app distinguishes starting, healthy, stopped, and unhealthy services. Its settings include model downloads and selection, a custom-model path, performance benchmarks, VS Code integration, copyable/exportable logs, and launch-at-login.

Read the complete Reticle MLX guide →

Use the VS Code extension with an existing endpoint

If you already have a compatible endpoint, install the extension directly:

code --install-extension roboalchemist.reticle

It is also available from Open VSX and as a VSIX on the latest GitHub release.

Open the Reticle target in the VS Code Activity Bar, then:

  1. Enter the endpoint base URL.
  2. Choose a model reported by /v1/models, or enter its ID manually.
  3. Select the matching FIM format.
  4. Save, click Check health, and use Try in editor.

The panel also exposes autocomplete state and live extension logs. Advanced options remain available through Open all settings.

Read the configuration guide →

Provider and setup guides

Reticle MLX is the easiest Mac setup, but the extension can use any compatible provider:

  • Reticle MLX
  • MTPLX
  • Ollama
  • llama.cpp
  • OMLX
  • LM Studio
  • Remote hosted endpoint

A server can return HTTP 200 while ignoring the suffix or wrapping the answer in chat prose. Verify the exact model and server together before relying on it:

Run model compatibility testing →

Documentation

  • How completion works
  • Configuration
  • Provider setup
  • Model compatibility testing and validated models
  • Troubleshooting
  • Reticle MLX for macOS
  • Development
  • Releasing
  • Changelog

License

MIT. The optional context engine is an original lightweight implementation; no third-party context-retrieval code is vendored.

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
  • Your Privacy Choices
  • Consumer Health Privacy
© 2026 Microsoft