cdv

azank1/cdv
★ 4 stars Python AI/LLM Updated today
CDV — the judge for AI coding agents. Every step your agent claims is scored through two channels (deterministic + LLM critic, stricter wins), with Bayesian stop logic, a git-stamped audit trail, and per-project memory. MCP server + VS Code extension. Local-first, no API key.
View on GitHub → 🔍 Audit Wallet Slippage →

Quick Install

Copy the config for your editor. Some servers may need additional setup — check the README.

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "cdv": {
      "command": "uvx",
      "args": [
        "cdv"
      ]
    }
  }
}

Or install with pip: pip install cdv

README Excerpt

**CDV — the judge for AI coding agents.** **Your agent said it works. CDV is the receipt.** Cursor, Copilot, Claude Code — every agent ends a task with "done, all tests pass." Sometimes that's true. CDV sits between your IDE agent and the model, scores each step the agent submits through **two independent

Tools (20)

CDV_DBCDV_MODELCDV_PROJECTCDV_PROVIDEROLLAMA_HOSTOPENROUTER_API_KEYcdv_analyze_promptcdv_classify_taskcdv_context_clearcdv_context_historycdv_dag_compilecdv_dag_mergecdv_dag_readycdv_dag_statuscdv_dag_submitcdv_feedbackcdv_gaugecdv_interceptcdv_list_taskscdv_loop_end

Topics

agentic-codingai-agentsclaude-codecode-reviewcursorllmmcpmodel-context-protocolverificationvscode-extension