mcp-local-rag
🥈SilverLocal-first RAG server for developers using MCP. Enables semantic and keyword search for code and technical docs. Fully private, zero setup. Ideal for operations teams needing secure, local search capabilities.
327190Updated 2mo ago
Intermediate30min to implementautomation
Saves ~150 min per use
Quick InstallView Source
git clone https://github.com/shinpr/mcp-local-rag.gitWorks with:
Claude
Overview
About This Skill
mcp-local-rag is a local RAG (retrieval-augmented generation) server that runs entirely on your machine without sending data to external APIs. It combines semantic vector search with keyword matching to find exact technical terms like error codes and function names alongside conceptual matches. The tool supports PDF, DOCX, TXT, and Markdown files, intelligently chunks documents by meaning rather than character count, and can optionally generate captions for visual content in PDFs. Setup requires only one command—no Docker, Python, or server configuration—and works offline after the initial model download. Developers, operations teams, and technical writers benefit from secure local search over code repositories, API documentation, research papers, and internal technical specs.
How to Use
["Install the MCP server locally using the provided package manager command (e.g., `npm install @modelcontextprotocol/local-rag`).","Initialize the RAG index by running `mcp-local-rag --index /path/to/codebase` to scan your project files.","Use the prompt template above in your AI assistant (e.g., Claude/ChatGPT) by replacing [TOPIC] and [CRITERIA] with your specific query.","For advanced searches, include file path filters (e.g., `/src/api/**`) or line number constraints in [CRITERIA].","Refresh the index periodically with `mcp-local-rag --reindex` to keep results up-to-date. For large codebases, schedule this as a nightly task."]
Use Cases
Search internal API documentation and technical specifications securely
Index code repositories and find relevant functions or patterns
Query research papers and technical papers with semantic understanding
Search error codes and logs with keyword boost for exact matches
Tags
Setup & Installation
Quick Install
No install command available. Check the GitHub repository for manual installation instructions.
Alternative Install (Git Clone)
git clone https://github.com/shinpr/mcp-local-ragRequirements
- Claude Code or compatible AI agent
- Works with: Claude
Quick Start Guide
1
Install the Skill
Copy the install command above and run it in your terminal.
2
Open Your AI Agent
Launch Claude Code, Cursor, or your preferred AI coding agent.
3
Try It Out
Use the prompt template or examples below to test the skill.
4
Customize
Adapt the skill to your specific use case and workflow.
Usage Examples
Prompt Template
Use the local RAG server to search for [TOPIC] in the codebase or technical documentation. Focus on finding relevant snippets, functions, or concepts that match [CRITERIA]. Provide the results in a structured format with file paths, line numbers, and a brief explanation of relevance. Example: 'Search for 'authentication middleware' in the /src/api directory and highlight functions that handle JWT validation.'
Example Output
```
**Search Results for 'authentication middleware' in /src/api/**
1. **File**: /src/api/middleware/auth.js
**Line**: 42-67
**Snippet**:
```javascript
const jwt = require('jsonwebtoken');
const authenticateToken = (req, res, next) => {
const token = req.headers['authorization']?.split(' ')[1];
if (!token) return res.sendStatus(401);
jwt.verify(token, process.env.JWT_SECRET, (err, user) => {
if (err) return res.sendStatus(403);
req.user = user;
next();
});
};
```
**Relevance**: Implements JWT-based authentication middleware used in 85% of API routes.
2. **File**: /src/api/routes/userRoutes.js
**Line**: 12-18
**Snippet**:
```javascript
router.get('/profile', authenticateToken, async (req, res) => {
const user = await User.findById(req.user.id);
res.json(user);
});
```
**Relevance**: Demonstrates the middleware in action for protected routes.
**Total Matches**: 12 files found. Top 2 shown above.
```
*Generated using local RAG server (MCP) on 2024-05-20. Indexed files: 478, total tokens: 1,245,890.*Apply to these tools
Browse all toolsIronCalc
IronCalc is a spreadsheet engine and ecosystem
Microsoft Teams
Get more done every day with Microsoft Teams – powered by AI
ServiceNow
Enterprise workflow automation and service management platform
GPT for work
Automate your spreadsheet tasks with AI power
Respell
Agentic AI Workflow platform
Notion
Connected workspace for docs, wikis, and projects
Compatible MCP servers
Browse all MCP servers
Find the right skills for your stack
Take a free 3-minute scan and get personalized AI skill recommendations.
Take free scan