Open Source MCP Servers
Model Context Protocol (MCP) servers connect AI assistants and agents to external tools, data sources, and services through a shared interface. They can expose capabilities such as searching files, querying databases, accessing web content, or controlling applications. By standardizing how these connections are described and used, MCP servers reduce the need for custom integrations and help assistants work with information beyond their built-in knowledge.
56 repositories · updated October 16, 2026
ragtime: Self-Host a RAG API and MCP Server
Ragtime connects local knowledge and tools to LLM clients through a self-hosted, OpenAI-compatible RAG API, MCP server, and built-in chat UI. It supports FAISS and PostgreSQL pgvector indexes, plus hosted and local model providers.
ActiveHealth 51/100
Koshi: Give AI Coding Agents Local Memory and Retrieval
Koshi is a local-first MCP server for giving AI coding agents durable project memory, code retrieval, and shared team decisions. It suits teams that want context to persist across sessions and machines without relying on cloud services or API keys.
QuietHealth 24/100
SocratiCode: Give AI Agents Searchable Codebase Context
SocratiCode is a local-first MCP server and plugin that helps AI agents explore large codebases with semantic search, dependency graphs, and symbol-level impact analysis. It also indexes project context such as API specs and database schemas.
ActiveHealth 75/100
lilbee: Run Local AI and Search Your Files
lilbee is a local AI model manager and search engine for files, code, and crawled websites. It offers cited answers through a terminal app, CLI, MCP server, REST API, and Python library.
ActiveHealth 50/100
ksail: Create and Operate Kubernetes Clusters
KSail is a Go-based toolkit for creating and operating Kubernetes clusters across local, nested, and cloud providers. It brings provisioning, GitOps, secrets, cluster operations, and AI interfaces into one tool, for developers and platform teams who want a unified workflow.
ActiveHealth 81/100
meshtastic-mcp: Connect AI Agents to Meshtastic Devices
A Python MCP server and agent skills for discovering, operating, observing, and testing Meshtastic devices and apps. Its portable core works over serial or TCP, with optional capabilities for firmware builds, mobile-app testing, replay, and hardware benches.
NewHealth 62/100
browser-rs-mcp: Share Chrome with AI Agents
A Rust MCP server that lets multiple AI agents control separate tabs in one shared Chrome profile. It provides browser automation through Chrome DevTools Protocol without requiring a Node.js runtime.
NewHealth 50/100
dcc-mcp-blender: Control Blender with AI Agents
An embedded MCP server that lets AI agents operate Blender through typed tools for modeling, UVs, rigging, materials, animation, and rendering. Suited to creators and pipeline teams seeking structured, cross-DCC automation.
NewHealth 61/100
open-index: Build Searchable Context Graphs for AI Agents
Open Index is a Python tool for building structured, searchable domain knowledge that agents can retrieve and maintain. It suits teams that need validated context across files, connectors, and MCP-enabled agents.
NewHealth 45/100
ferret-mcp: Analyze Codebases Through MCP
Ferret MCP gives AI assistants static codebase analysis and optional LLM-powered reports about architecture, dependencies, patterns, and APIs. It suits developers exploring unfamiliar repositories, but its quiet, single-contributor project status is worth weighing before relying on it in a team workflow.
QuietHealth 20/100
DeclarAgent: Run Auditable Workflows for AI Agents
DeclarAgent runs YAML-defined workflows through shell commands, built-in actions, and HTTP requests. Its validation, dry-run, approval gates, and MCP integration suit teams that want agents to invoke controlled CLI workflows.
QuietHealth 17/100
agenticschema: Turn Structured Web Data Into Agent Tools
AgenticSchema converts Schema.org data already present in web pages into tools for AI agents. It supports browser WebMCP registration and a Node MCP server, without requiring a new backend or page API.
NewHealth 44/100