⚔️ Excalibur: Peak Frontier Intelligence On Demand

Multi-model council consensus and adversarial critique engine polling Claude 3.7 Sonnet/Opus, OpenAI o3-mini/o1, Google Gemini 2.0 Pro, and DeepSeek R1.

---

🌟 Why Excalibur?

Single AI models suffer from vendor bias, blind spots, and hallucinations. Excalibur queries 4 leading frontier reasoning models in parallel, debates their discrepancies, and produces a harmonious consensus deliverable:

1. Pre-Execution Blueprint (Planning): Validate architecture, roadmaps, and risk models before building.
2. Post-Execution Verification (Critique): Deep code security audits, contract redlines, edge-case vulnerability detection.
3. Pure Pay-Per-Use: Upload files, get dynamic token volume estimates, and pay per audit ($1.90 - $14.90).
4. Native MCP Server: Plug directly into Claude Desktop & Cursor with 1 click.

---

🛠️ Installation & MCP Setup

Claude Desktop Configuration

{
  "mcpServers": {
    "excalibur": {
      "command": "npx",
      "args": ["-y", "@smithery/cli", "install", "excalibur", "--client", "claude"]
    }
  }
}

CLI Usage

python backend/cli.py audit path/to/code_or_contract.ts --type code_security --stage critique