Working with VS Code

WS Code Review is a Visual Studio Code extension for offline, privacy-first AI code reviews tailored to F# and WebSharper.

• Runs entirely on your machine via Ollama.
• Uses a local coder model (default: qwen2.5-coder:7b-instruct).
• No code is sent to cloud services.

Links

• Marketplace: https://marketplace.visualstudio.com/items?itemName=souvanxay.ws-code-review
• GitHub + README: https://github.com/Got17/ai-code-review-vscode

Key features

• Selection-based reviews with streamed markdown and a diff preview.

• Apply mode:

  • Normal files: applies an improved full file — primarily the selected region changes, but related code may also be adjusted if required (e.g., imports, types, call sites).

  • Large files (≥ 600 lines): applies changes to the selected region only.

• Model switcher (in webview + command). Default model is qwen2.5-coder:7b-instruct.

• RAG pill (ON/OFF) in the top bar of the webview to enrich reviews with bundled references.

• Shadow Git snapshots (optional): accepted suggestions are snapshot-committed to a private repo in extension storage—your real repo is untouched.

• AI preferences: store style/constraints (e.g., "functional style", "no renames") and show them in the webview.

Quick start

1. Install from the Marketplace (link above).

Here’s the updated Quick start → Step 2 you can paste in:

2. Install and start Ollama:

   • Download and install Ollama for your OS (from the official site).

   • Pull the coder model:

   ollama pull qwen2.5-coder:7b-instruct

   • Run the Ollama Server

   ollama serve

   By default, Ollama serves at http://localhost:11434, open it in your browser to verify (you should see "Ollama is running").

Tip: If you prefer another local model, pull it first (e.g., ollama pull <model:tag>) and run WS Code Review: Change Ollama Model on Command Palette to switch before running Show Suggestion.

3. Open an .fs file, select code, then run:

   • WS Code Review: Show Suggestion
   • Shortcut: Ctrl+Alt+R (Windows/Linux) or Ctrl+Cmd+R (macOS)

4. Review the streamed suggestion and diff → Accept or Reject.

5. (Optional) Click the RAG pill (ON/OFF) and press Refresh to re-run.

6. (Optional) Enable Shadow Git in settings to keep private snapshot history.

Commands

• WS Code Review: Show Suggestion — run review on the current selection.
• WS Code Review: Change Ollama Model — pick a local model or enter a tag.
• WS Code Review: Set / Show / Clear AI Preferences — manage reviewer hints.
• WS Code Review: Show Shadow Git History (Current File) — browse snapshot commits.
• WS Code Review: Clear Shadow Git History — purge the private snapshot repo.

See the GitHub README for the full command list, options, and troubleshooting.

Settings (high level)

• wsCodeReview.rag.enable — toggle retrieval-augmented context (also via the RAG pill).
• wsCodeReview.git.enable — enable private Shadow Git snapshotting on Accept.

Endpoint and model are configurable via Change Ollama Model. Default endpoint: http://localhost:11434/api/generate Default model: qwen2.5-coder:7b-instruct

Privacy

This extension communicates only with a local Ollama server. No source code or preferences are sent to external/cloud services. Shadow Git history is stored in the extension’s private storage and can be cleared at any time.