loop-llm

azank1/loop-llm

★ 2 stars Python Code/Dev Tools Updated today

MCP server that observes every prompt, scores quality in real time, and closes the loop with iterative refinement. Built on FastMCP, SQLite, and Bayesian priors — no extra LLM required.

View on GitHub → Try with Claude — $10 free →

Quick Install

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

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "loop-llm": {
      "command": "uvx",
      "args": [
        "loop-llm"
      ]
    }
  }
}

Or install with pip: pip install loop-llm

README Excerpt

**A Bayesian MCP sidecar for your IDE agent** — observe prompts, refine outputs, and stop agent loops using externally verified scores. > The underlying CLI and tool API ship as `loopllm` (the original name). PromptLoop is the project brand. Current release: **v0.7.0**. --- PromptLoop sits between you and Cursor, VS Code Copilot, or any MCP client. It does

Tools (20)

LOOPLLM_DBLOOPLLM_MODELLOOPLLM_PROVIDEROLLAMA_HOSTOPENROUTER_API_KEYloopllm_analyze_promptloopllm_classify_taskloopllm_context_clearloopllm_context_historyloopllm_feedbackloopllm_gaugeloopllm_interceptloopllm_list_tasksloopllm_loop_endloopllm_loop_startloopllm_loop_statusloopllm_loop_steploopllm_plan_deleteloopllm_plan_listloopllm_plan_next

loop-llm

Quick Install

README Excerpt

Tools (20)

Related Code/Dev Tools Servers