Back to Apps

DevRag
Supports UIby tomohiro-owada
Lightweight local RAG system for Claude Code using markdown vector search and multilingual embeddings.
0 stars
Works in:claude
Exposes:Tools
What it does
DevRag is a local Retrieval-Augmented Generation (RAG) system designed specifically for developers. It indexes local markdown files into a vector database, allowing AI agents to find relevant documentation snippets via semantic search instead of reading entire files, significantly reducing token usage.
Tools
search: Perform semantic vector search for natural language queries.index_markdown: Add a specific markdown file to the vector index.list_documents: Retrieve a list of all currently indexed documents.delete_document: Remove a specific document from the index.reindex_document: Refresh the index for a modified markdown file.
Installation
After downloading the binary to /usr/local/bin/devrag, add this to your claude_desktop_config.json:
{
"mcpServers": {
"devrag": {
"command": "/usr/local/bin/devrag"
}
}
}
Supported hosts
Claude Code, Claude Desktop.
Choose your AI client and follow the steps below.
Claude Desktop
{
"mcpServers": {
"devrag": {
"command": "/usr/local/bin/devrag"
}
}
}





