Loopy MCP Server
Integrates Loopy's AI agent loop library with MCP-compatible clients like LM Studio, enabling discovery, analysis, crafting, and execution of bounded agent workflows.
README
Loopy MCP Server
A Model Context Protocol (MCP) server that brings Loopy functionality to LM Studio and other MCP-compatible clients.
What is Loopy?
Loopy is a library of practical AI-agent loops and tools for discovering, adapting, crafting, and running repeatable agent workflows. A loop is a bounded, feedback-driven workflow that helps agents learn from results and take the next useful step.
From Forward Future: https://signals.forwardfuture.com/loop-library/
Why Loopy MCP Server?
LM Studio doesn't have a skill system (yet), so this MCP server bridges the gap by exposing Loopy's core functionality through the MCP protocol. Now you can use Loopy directly in LM Studio to:
- Search and discover published loops from the Loop Library
- Get recommendations for your specific tasks
- Analyze loops for weaknesses and improvements
- Craft new custom loops through guided questions
- Save loops to your project for reuse
- Discover repeated patterns in your codebase
- Execute loops with bounded passes and track results
Features
✅ Search the Loop Library - Find published loops by keyword
✅ Get Recommendations - Discover loops for your specific task
✅ Analyze Loops - Audit for weaknesses, stopping conditions, verification
✅ Craft Loops - Interactive questions to design custom loops
✅ Save Loops - Store loops in project's LOOPS.md
✅ Load Project Loops - Reuse saved loops
✅ Discover Patterns - Find repeated patterns in your codebase
✅ Run Loops - Execute with bounded passes and get receipts
✅ Live Catalog - Connects to the latest Loop Library
Installation
Prerequisites
- Node.js 20+
- npm or yarn
Setup
git clone https://github.com/hotkid61/Loopy-MCP-Server.git
cd Loopy-MCP-Server
npm install
Usage
Start the Server
npm run server
The server will start listening for JSON-RPC requests on stdin/stdout.
For LM Studio
Add this to your LM Studio MCP configuration:
{
"mcpServers": {
"loopy": {
"command": "npm",
"args": [
"--prefix",
"/path/to/Loopy-MCP-Server",
"run",
"server"
]
}
}
}
Replace /path/to/Loopy-MCP-Server with your actual installation path.
Available Tools
search_loops
Search the Loop Library catalog for loops by keyword
query: string (required) - Search keywords
limit: number (optional) - Max results (default: 5)
get_loop_recommendations
Get loop recommendations based on your task
task: string (required) - Describe what you want to accomplish
get_loop_details
Get full details of a specific loop
loop_id: string (required) - Loop ID or name
analyze_loop
Audit a loop for weaknesses and get suggestions
loop_prompt: string (required) - The loop instructions to analyze
craft_loop
Get interview questions to help design a custom loop
save_loop
Save a loop to your project's LOOPS.md
name: string (required) - Loop name
description: string (required) - One-sentence description
prompt: string (required) - The loop instructions
source: string (optional) - Source URL if adapted from published loop
load_project_loops
Load all loops saved in your project's LOOPS.md
discover_patterns
Analyze your codebase for repeated patterns
run_loop
Execute a loop with bounded passes
loop_prompt: string (required) - The loop to execute
max_passes: number (optional) - Max passes (default: 10)
loop_name: string (optional) - Loop name for tracking
list_categories
List all categories in the Loop Library
File Structure
Loopy-MCP-Server/
├── loopy-mcp-server.ts # Main MCP server
├── catalog-client.ts # Loop Library catalog integration
├── loopy-tools.ts # Core Loopy functionality
├── loop-executor.ts # Loop execution and tracking
├── package.json # Dependencies
├── tsconfig.json # TypeScript config
├── .gitignore # Git ignore rules
└── README.md # This file
Examples
Find a loop for your task
Request: "Find a loop for improving test coverage"
- Searches the catalog
- Returns top 3 recommendations
- Shows use cases, prompts, verification steps
Analyze an existing loop
Request: "Audit this loop and suggest improvements: [paste loop]"
- Checks for clear stopping conditions
- Verifies acceptance criteria
- Identifies gaps
- Suggests improvements
Save and reuse project loops
Request: "Save this loop to the project"
- Saves to LOOPS.md
- Available in future sessions
- Can be exported to Loop Library
Run a loop with tracking
Request: "Run the Improve Test Coverage loop in this project"
- Executes in bounded passes
- Tracks actions and evidence
- Provides receipt with outcome
- Suggests improvements via debrief
How Loops Work
A good loop answers four questions:
- What are you trying to accomplish? - Clear goal
- How will you know it worked? - Acceptance criteria
- What should you do with what you learn? - Feedback cycle
- When should you stop? - Exit condition
Example:
Find the slowest page, make one focused improvement, and measure again. Keep the change only if it helps. Repeat until every page meets the target or another pass stops producing improvement.
Integration with Forward Future
This server connects to the official Loop Library maintained by Forward Future:
- Catalog: https://signals.forwardfuture.com/loop-library/
- Repository: https://github.com/Forward-Future/loopy
- License: MIT
Loops are published with quality checks and can be submitted to the catalog from your project.
Features in Detail
Loop Discovery
Analyzes your codebase for repeated engineering patterns and turns strong candidates into bounded loops.
Loop Adaptation
Takes a published loop and tailors it to your tools, limits, schedule, and definition of success.
Loop Execution
Runs loops in bounded passes, applies acceptance checks, and returns evidence-backed receipts.
Loop Debrief
Analyzes run results and recommends the smallest justified improvement.
Publication
Validates, checks catalog overlap, and prepares loops for publication in Loop Library.
Notes
- Catalogs are cached for 1 hour to reduce network requests
- Loop execution is simulated in basic mode (can be extended with real execution)
- LOOPS.md stores project-local loops for reuse
- LOOPS_RUNS.md tracks execution receipts for analysis
- All operations are non-destructive unless explicitly requested
License
MIT
Built With
- TypeScript
- Node.js
- MCP Protocol
- Loop Library (Forward Future)
Support
For issues with Loopy itself: https://github.com/Forward-Future/loopy For MCP server issues: Create an issue in this repository
Ready to bring Loopy to LM Studio! 🚀
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.