Knowledge Rag MCP Server

Drop your PDFs, markdown, code, notebooks — **1800+ files, 39K chunks, indexed in under 3 minutes.**

Local serverstdioPython

What is the Knowledge Rag MCP server?

Knowledge Rag MCP server exists for a simple reason — assistants are far more useful when they can act on Knowledge Rag directly instead of describing what you should do. Drop your PDFs, markdown, code, notebooks — 1800+ files, 39K chunks, indexed in under 3 minutes..

What the assistant can call

Once Knowledge Rag is connected, these are the calls the assistant has available:

  • v4.0.0Enterprise concurrent access: SSE/HTTP transport (1 server → N clients), thread-safe shared state, optional rate limiting + Prometheus metrics
  • v3.9.0Quality Gate activated: 35+ automated PR checks across 7 pillars (Security, Stability, Memory Leak, Versatility, Scalability, Versioning
  • v3.8.1 — Critical hotfix: loud-fail embeddings (no more silent zero-vector corruption); Windows CI flake erradicated (HF_HUB_OFFLINE + shell:bash + atexit
  • v3.8.0 — Lazy-load embeddings, opt-in single-instance guard, version sync across PyPI/NPM/Docker
  • v3.6.0 — Multi-language code parsing (C/C++/JS/TS/XML), NPM wrapper, Docker image, automated release pipeline
  • v3.5.2 — CUDA DLL auto-discovery from pip packages, graceful GPU→CPU fallback, explicit CPU provider (no CUDA noise when gpu: false), BASE_DIR resolution fix
  • v3.5.1 — Remove Python <3.13 upper bound — 3.13 and 3.14 now supported
  • v3.5.0 — Optional GPU acceleration, supported formats table, full README rewrite
  • v3.4.3 — MCP stdout save/restore fix (v3.4.2 broke JSON-RPC responses)
  • v3.4.0 — Persistent model cache, exclude patterns, Jupyter Notebook parser, inotify resilience, MetaTrader support
  • Optional — * NVIDIA GPU + CUDA 12 for accelerated embeddings (see GPU Acceleration below)
  • Pillar — What it enforces

Setting it up

The server ships on npm as knowledge-rag, so your MCP client can launch it on demand — there is no separate build step. Add the server block to your client's configuration, restart it, and the tools register themselves.

Configuration and credentials

You will need one environment variable: KNOWLEDGE_RAG_SINGLE_INSTANCE. Keep credentials in your client's env block or a secrets manager rather than in a file you might commit.

  • Python 3.11+ - Claude Code CLI - …or any other MCP client (Claude Desktop, Cursor, VS Code, Antigravity, opencode, Windsurf) — see Use with other MCP clients - ~200MB disk for model cache (auto-downloaded on first run) - Optional: NVIDIA GPU + CUDA 12 for accelerated embeddings (see GPU Acceleration below)

Choosing this one

This sits in the cloud and infrastructure group, where several servers overlap in what they claim to do but differ sharply once you actually set them up. Knowledge Rag's toolset — v4.0.0, v3.9.0, v3.8.1 and 11 more — is a fair guide to whether it matches your workflow. It is maintained by lyonzin; worth a glance at recent repository activity before you build anything load-bearing on it.

This entry was verified against Knowledge Rag's own documentation before publication; SyncDev keeps the directory reviewed rather than auto-generated.

Before you rely on it

  • It runs with your machine's permissions. That is convenient and also the reason to think about what you point it at before you approve a tool call.
  • With 14 tools registered it takes up a noticeable share of the context window; turn it off in projects that never touch Knowledge Rag.
  • MCP clients confirm each tool call by default. Leave that on until you have watched what the knowledge rag mcp server does with a few real requests.

Available tools

ToolWhat it does
v4.0.0**Enterprise concurrent access**: SSE/HTTP transport (1 server → N clients), thread-safe shared state, optional rate limiting + Prometheus metrics, ChromaDB WAL mode, --transport CLI
v3.9.0**Quality Gate** activated: 35+ automated PR checks across 7 pillars (Security, Stability, Memory Leak, Versatility, Scalability, Versioning, Quality) + nightly resilience suite (chaos, soak, determinism, mutation)
v3.8.1Critical hotfix: loud-fail embeddings (no more silent zero-vector corruption); Windows CI flake erradicated (HF_HUB_OFFLINE + shell:bash + atexit wrapper)
v3.8.0Lazy-load embeddings, opt-in single-instance guard, version sync across PyPI/NPM/Docker
v3.6.0Multi-language code parsing (C/C++/JS/TS/XML), NPM wrapper, Docker image, automated release pipeline
v3.5.2CUDA DLL auto-discovery from pip packages, graceful GPU→CPU fallback, explicit CPU provider (no CUDA noise when gpu: false), BASE_DIR resolution fix for editable installs
v3.5.1Remove Python <3.13 upper bound — 3.13 and 3.14 now supported
v3.5.0Optional GPU acceleration, supported formats table, full README rewrite
v3.4.3MCP stdout save/restore fix (v3.4.2 broke JSON-RPC responses)
v3.4.0Persistent model cache, exclude patterns, Jupyter Notebook parser, inotify resilience, MetaTrader support
Optional* NVIDIA GPU + CUDA 12 for accelerated embeddings (see [GPU Acceleration](#gpu-acceleration) below)
PillarWhat it enforces
FormatExtension
Markdown.md

How to install the Knowledge Rag MCP server

{
  "mcpServers": {
    "knowledge-rag": {
      "command": "npx",
      "args": ["-y", "knowledge-rag"],
      "env": {
        "KNOWLEDGE_RAG_SINGLE_INSTANCE": "your-value"
      }
    }
  }
}

Add to claude_desktop_config.json, then restart Claude Desktop.

Configuration

  • Python 3.11+ - Claude Code CLI - …or any other MCP client (Claude Desktop, Cursor, VS Code, Antigravity, opencode, Windsurf) — see Use with other MCP clients - ~200MB disk for model cache (auto-downloaded on first run) - Optional: NVIDIA GPU + CUDA 12 for accelerated embeddings (see GPU Acceleration below)
VariableDescriptionRequired
KNOWLEDGE_RAG_SINGLE_INSTANCEConfiguration value read at startup.Optional

Example prompts to try

  • Use Knowledge Rag to v4.0.0.
  • Use Knowledge Rag to v3.9.0.
  • Use Knowledge Rag to v3.8.1.

Frequently asked questions

It connects Knowledge Rag to MCP-compatible AI assistants such as Claude and Cursor, exposing 14 tools (v4.0.0, v3.9.0, v3.8.1, and more) that the assistant can call on your behalf. Instead of copying data back and forth by hand, the assistant works with Knowledge Rag directly.