Back to Apps

Code RAG MCP
by Mirrdhyn
Semantic code search for AI agents. Replace grep with understanding-based search using local embeddings via LM Studio.
0 stars
Works in:zed
Exposes:Tools
What it does
Code RAG MCP transforms how AI agents interact with codebases by replacing traditional text-based search (grep/ripgrep) with semantic, understanding-based retrieval. It enables agents to find code by concept or intent rather than exact keyword matches.
Tools
semantic_code_search: The primary search tool for finding code by concept.find_similar_code: Locates code snippets similar to a provided example.explain_code_with_context: Provides an explanation of a file while considering its dependencies.index_codebase: Indexes a directory to enable semantic search.get_index_stats: Returns the current status and statistics of the index.
Installation
{
"mcpServers": {
"code-rag": {
"command": "/usr/local/bin/code-rag-mcp",
"env": {
"LM_STUDIO_URL": "http://localhost:1234/v1"
}
}
}
}
Supported hosts
Confirmed for Claude Desktop, Claude Code, and Zed.
Quick install
go build -o code-rag-mcp && sudo cp code-rag-mcp /usr/local/bin/Information
- Pricing
- free
- Published
- 6/13/2026
- stars
- 0
Categories
Choose your AI client and follow the steps below.
zed
{
"context_servers": {
"code-rag": {
"source": "custom",
"command": {
"path": "/usr/local/bin/code-rag-mcp",
"args": [],
"env": {
"LM_STUDIO_URL": "http://localhost:1234/v1"
}
}
}
}
}Claude Desktop
{
"mcpServers": {
"code-rag": {
"command": "/usr/local/bin/code-rag-mcp",
"env": {
"LM_STUDIO_URL": "http://localhost:1234/v1"
}
}
}
}





