Judges Panel — VS Code Extension

Install one extension. Get 35 specialized judges reviewing every file you save — inline diagnostics, quick-fix patches, @judges chat participant for Copilot, and automatic MCP server integration so Copilot and other LLMs can act as 35 expert reviewers with zero configuration.

Features

@judges chat participant — type @judges in Copilot Chat to review the active file, run security checks, or auto-fix findings. Copilot also auto-routes when you mention "judges panel review", "judges evaluation", etc.
Language Model tool — judges_evaluate is registered as an LM tool so Copilot can automatically discover and invoke Judges evaluation even without the @ prefix
Evaluate on save — automatically runs the full 35-judge tribunal when you save a file
Inline diagnostics — findings appear as squiggly underlines with severity-colored markers
Quick-fix code actions — click the lightbulb to apply auto-fix patches directly
Auto-configured MCP server — the 35 expert-persona LLM prompts are registered with VS Code automatically; Copilot can use them immediately
Workspace evaluation — evaluate all supported files with progress reporting
Status bar — one-click evaluation from the status bar shield icon
Configurable — choose presets, severity filters, and specific judges

Chat Participant — `@judges`

Talk to the Judges Panel directly in Copilot Chat:

Command	What it does
`@judges`	Review the active file with all 35 judges
`@judges /review`	Same as above
`@judges /security`	Security-focused review only
`@judges /fix`	Auto-fix all fixable findings
`@judges /help`	Show available commands

You can also ask naturally — Copilot auto-routes these to @judges:

"Review this file with judges"
"Run a judges panel evaluation"
"Check this code for security issues"

Supported Languages

TypeScript, JavaScript, Python, Go, Rust, Java, C#, C++

Commands

Command	Description
`Judges: Evaluate Current File`	Run the full tribunal on the active file
`Judges: Auto-Fix Current File`	Apply all available auto-fix patches
`Judges: Evaluate Workspace`	Evaluate all supported files in the workspace
`Judges: Clear Diagnostics`	Remove all Judges diagnostics
`Judges: Show Results Panel`	Open the Judges results view
`Judges: Configure MCP Server`	Write MCP server definition to `.vscode/mcp.json` (manual alternative to auto-config)

Configuration

Setting	Default	Description
`judges.evaluateOnSave`	`true`	Automatically evaluate on file save
`judges.preset`	`""`	Named preset: strict, lenient, security-only, etc.
`judges.minSeverity`	`"medium"`	Minimum severity to display
`judges.enabledJudges`	`[]`	Restrict to specific judge IDs
`judges.debounceMs`	`1000`	Debounce delay for on-save evaluation

Installation

From VS Code Marketplace (recommended)

Search for "Judges Panel" in the Extensions view (Ctrl+Shift+X) or run:

code --install-extension kevinrabun.judges-panel

That's it — Layer 1 (deterministic evaluation) and Layer 2 (MCP server for LLM-powered review) are both enabled automatically.

From Source (Development)

cd vscode-extension
npm install
npm run compile
# Press F5 in VS Code to launch Extension Development Host

From VSIX

cd vscode-extension
npx @vscode/vsce package
code --install-extension judges-panel-*.vsix

How It Works

The extension provides both layers of the Judges evaluation system with zero configuration:

Layer 1 — Deterministic Evaluation (instant, local)

Runs locally in the extension host — no network calls, no API keys, instant results:

Pattern matching against 400+ security, quality, and compliance rules
AST analysis via tree-sitter WASM for 8 languages
Auto-fix patches for 47+ common issues

Layer 2 — LLM-Powered Deep Review (enabled automatically)

The extension automatically registers the Judges MCP server with VS Code on activation. Copilot and other LMs see all 35 expert-persona prompts immediately — no .vscode/mcp.json, no npx commands, no manual setup of any kind.

What this means in practice: ask Copilot to "review this file with the Judges cybersecurity expert" and it can — because the MCP server is already running.

Chat Participant & LM Tool (zero-friction Copilot integration)

The extension registers an @judges chat participant and a judges_evaluate Language Model tool. Together these ensure Copilot can discover and invoke the Judges Panel automatically:

@judges — users can type @judges /review in chat, or simply mention "judges panel review" and Copilot routes the request via VS Code's disambiguation
judges_evaluate tool — registered via vscode.lm.registerTool, Copilot may auto-invoke this tool when the user mentions "judges", "code review", or "evaluate code"

If you prefer explicit workspace-level config (e.g., to share with teammates who don't have the extension), run Judges: Configure MCP Server to write the server definition to .vscode/mcp.json.

License

MIT

Judges Panel

Kevin Rabun