MCP servers that ship as code — a repository or a package you install and run locally (usually over stdio). They have no hosted endpoint, so we cannot probe or grade them: instead each one shows its repository, install command, license, stars and when it was last updated.
Collected from GitHub repositories and npm, PyPI, Cargo, Go, Maven, NuGet and RubyGems packages tagged as MCP servers, the official MCP Registry, Docker's MCP catalog, ModelScope, Baidu MCP World, Cline and other public directories; one entry per repository. A server that also offers a hosted endpoint is listed under Remote instead. These entries are not counted in our reachability, verification or readiness statistics.
-
mcp-llm-routergithub.com/groxaxo/mcp-llm-routerupdated this year
—
★ 3 -
Universal LLM router for AI coding tools. Works with Claude Code, Cursor, Codex, Gemini CLI, Copilot and more. Free-first fallback chain cuts costs 35-80%.
npm MIT ★ 130 -
mcp-llm-routergithub.com/ppikkuaho/mcp-llm-routerupdated this year
A multi-provider LLM router exposed as an MCP server: one agent loop over OpenAI and Gemini with retry, rate limiting, and dynamic tool schemas.
MIT -
LLM-Router-MCPgithub.com/Devatva24/LLM-Router-MCPupdated this year
MCP server that intelligently routes prompts across Claude, Gemini, and GPT-4o based on task type — minimising token cost without sacrificing quality.
npm -
llm-cost-router-mcpupdated this year
MCP server that advises LLM model routing & cost BEFORE the call — estimate_cost, recommend_model, should_offload_local. Pure logic, no API keys, $0 to run.
npm MIT -
Cheapest-LLM Router — a zero-dependency MCP server that routes any prompt to the cheapest model that can actually serve it: free models first, then the lowest-cost paid fallback. Returns USD/CNY cost, a fallback chain and plain-English reasoning. 4 tools, no npm install, calls no model itself.
MIT -
agent-router-mcpgithub.com/forge18/agent-router-mcpupdated this year
Intelligent MCP router for AI subagents with config-driven rules and LLM semantic tagging
MIT ★ 1 -
Smart LLM router MCP server. Route prompts to 20+ text models (Claude, GPT, Gemini, DeepSeek) with cascade strategy to cut costs 40-50%.
npm Apache-2.0 -
💎 Semantic Context Compressor, Live Token/USD Cost Calculator & Adaptive Model-Tier Router MCP Server — Cut LLM token bills by 85% across Claude Code, Cursor, VS Code, Codex & Antigravity.
MIT ★ 1 -
Official SDKs for the AI Stupid Level Smart Router (benchmark-driven LLM router) and Data API: TypeScript, Python, Vercel AI SDK provider, MCP server, LangChain, n8n.
npm MIT -
llm-localfirstgithub.com/shaxzodbek-uzb/llm-localfirstupdated this year
Local-first LLM router — keep sensitive data and bulk text labor on your own models; call the cloud only for the hard part. Fail-closed privacy routing + manager-worker delegation, MCP-native.
pypi MIT -
MCP server for the three.ws multi-provider LLM router: discover models and run chat completions through any of them.
Apache-2.0 ★ 9 -
freemodel-mcpgithub.com/yummy342/freemodel-mcpupdated this year
Use the best model for every task — automatically. Multi-provider LLM router for Claude Code. Picks the right model based on what you are doing: coding, writing, reasoning. 25+ platforms (DeepSeek, Qwen, GLM, Gemini, Claude, GPT), one API key. Save 40-70%
npm MIT -
Use the best model for every task — automatically. Multi-provider LLM router for Claude Code. Picks the right model based on what you are doing: coding, writing, reasoning. 25+ platforms (DeepSeek, Qwen, GLM, Gemini, Claude, GPT), one API key. Save 40-70%
npm MIT -
Delegate Claude Code work to cheaper and local models: an MCP server with a cost-aware router, privacy levels that fail closed, a local LLM node pool and an optional local orchestrator gateway.
MIT -
⚡ Universal Model Context Protocol (MCP) Server & Agentic Tool Router for Claude Code, Cursor, and Local LLMs — Built by Sarvesh Zenith (Dayal Infotech Services)
MIT -
marsgithub.com/asakur44/marsupdated this year
MARS — Model Adapter & Routing System. MCP-based router for multi-model LLM dispatch with sessions and budgets. Part of the Fr4ym + MARS + BCKS stack.
MIT -
⚡ Deterministic <15ms coding task router (Model Tier + Effort Level + Subagent Mode) that learns from real git diff survival. Zero LLM cost, 100% local SQLite.
-
Agentic LLM router that sends each request to a LoRA-tuned small model or a larger one by confidence, cutting cost and latency without losing quality. LangGraph, RAG, CNN, Groq and Gemini, with PII and prompt-injection guards. Docker, Prometheus, MCP server..
NOASSERTION ★ 3 -
MCP server exposing TP-Link Deco mesh routers (nodes, clients, Wi-Fi, blocklist, fixed IPs) as tools for LLM clients
pypi MIT -
Official 2026 LLM API Pricing Calendar, 272K/100K Context-Cliff Router, Chrome FinOps HUD Extension & Cursor MCP Server by Etuki.com
★ 1 -
Local-first GraphRAG over a markdown knowledge base: LightRAG, Ollama/Qwen, MCP server, LLM router, scheduled self-maintenance
MIT -
Self-hosted LLM operations gateway in one Go binary: OpenAI-compatible router, durable agent tracker, faithfulness evals + drift, traces, MCP server, console
Apache-2.0 ★ 2 -
LangGraph data-analyst agent: LLM router + ReAct over typed Pydantic tools, SQLite memory, CLI / Streamlit / MCP server; 95% routing accuracy
-
freeloadergithub.com/1977-code/freeloaderupdated this year
Claude Code MCP server + skill that offloads cheap, bulk subtasks to free local LLMs via a self-hosted freellmapi router. Save your Opus/Sonnet tokens.
MIT -
Semantic fork of camofox-mcp: an LLM router lets AI agents drive an anti-detection browser by intent instead of by refs and selectors.
MIT -
Multi-agent RAG on LangGraph: agents research, write and fact-check cited daily notes. Hybrid search (Postgres full-text + pgvector), an LLM router with fallbacks, evals in CI, an MCP server. FastAPI, Astro.
Apache-2.0 -
crosscheck-mcpactive
Multi-LLM MCP server: confer / debate / coordinate / audit / orchestrate across Anthropic + OpenAI + xAI + Gemini + Mistral + Groq + DeepSeek + Kimi + Qwen, with scoreboard-driven router, canary-leak detection, sandboxed shell verifiers, and a tier-aware
npm SEE LICENSE IN LICENSE -
Token‑efficient LLM router + context manager for n8n and Claude Code (MCP)—deployable locally or via remote HTTP bridges.
MIT ★ 1 -
unchained-mcpactive
Claude MCP Connector and Agent Skill for Unchained Code (Cost-Free LLM Coding Router & Optimization Engine)
npm MIT -
Self-hosted .NET 10 module monolith: infrastructure for AI coding agents - session digests, hybrid FTS5+vector memory search (RRF + rerank), task boards with a methodology engine, a fault-tolerant LLM router and a 100+ tool MCP server.
npm MIT ★ 1 -
Zero-trust AI control plane: strict stdio MCP server, MCP security gateway (JWT + RBAC), streaming PII guardrail, and a rate-limited failover LLM router, with an end-to-end demo and 348 tests.
-
🌱 LLAAB: Order out of chaos by turning unstructured information into structured, executable knowledge. Local-first TypeScript monorepo: YouTube ingestion pipeline, typed vault nodes, LLM router, agent orchestration, and MCP server.
-
customer-service-agent-carmit-haasgithub.com/CarmitHaas/customer-service-agent-carmit-haasupdated this year
LangGraph ReAct data-analyst agent over the Bitext customer-service dataset: query router, Pydantic tools, persistent memory + user profile, FastMCP server. Nebius LLMs.
MIT -
zelpiactive
ZelPi — Physical Intelligence Operating System. A self-contained CLI OS for autonomous robot fleets: kernel QoS router, model registry (VLM/VLA/LLM), MCP server, and an embedded simulation backend. Command: `npx zelpi`.
npm MIT -
The MCP-RAG project is an AI agent framework based on the Multi-Context Protocol that connects various data sources (MySQL, documents, RAG search) with LLMs. It supports HTTP API, SSE chat, an MCP router, batch workers, and a React + Tailwind frontend.
Apache-2.0 ★ 10 -
Global skill vault (~/.skill-proof/skills) and deterministic in-process skill router for AI coding agents & IDEs — zero routing LLM calls, sub-millisecond lexical + fuzzy + local vector routing, negation vetoes, and universal stdio MCP server.
MIT ★ 1 -
Vetono update data
93 agentic MCP tools + 49 specialist agents for every major AI CLI — self-learning, no API keys. Veto gives Claude Code, Codex CLI, Gemini CLI, Cursor, Windsurf, Zed, and JetBrains a council of specialist agents, persistent cross-platform memory, a self-learning router, CI/CD gates, and dependency-hallucination + decision-drift guards — all powered by your existing AI subscription. No separate API keys, no extra billing. **Highlights** - **Council debate** — multi-agent GREEN/YELLOW/RED verdicts on risky changes (LLM-backed via MCP Sampling, with deterministic fallback) - **Cross-platform memory & session handoff** — save on Claude, resume on Gemini or Codex; nothing needs re-explaining - **Self-learning router** — improves task routing from recorded outcomes - **Guards & gates** — CI gate, diff review, secrets scan, dependency-hallucination verification, decision-drift enforcement - **Compact mode** — advertise 9 tools instead of 93 and save ~16K context tokens; every tool stays callable **Links**: [GitHub](https://github.com/jigyasudham/veto) · [npm](https://www.npmjs.com/package/@jigyasudham/veto)
smithery -
concept-routergithub.com/prabhukirangit/concept-routerupdated this year
A zero-infra MCP server that turns an Open Knowledge Format (OKF) bundle into fast, lexical-first knowledge discovery for LLM agents — no vector DB, no embeddings by default.
Apache-2.0 -
mcp-brain-routergithub.com/mohan-n-swamy/mcp-brain-routerupdated this year
MCP connector for multi-backend LLM routing (DeepSeek, GLM, Codex) with native Claude Code orchestration
pypi MIT -
baiduan-routeractive
Route prompts across 9 LLM backends through a single MCP server — compare, optimize, and save. No lock-in, no vendor bias.
pypi Apache-2.0 -
MCP服务器,根据模式自动选择和调用不同的大语言模型
npm MIT -
Lightweight, single-binary, production-grade Rust LLM routing gateway & dashboard with health-aware failover, Claude Code compatibility, and MCP server.
Apache-2.0 ★ 4 -
Keep skill catalogs outside the main LLM context. Jev selects relevant skills through one read-only MCP tool.
MIT ★ 4 -
Graph-aware code context for CLI and MCP workflows
MIT ★ 3 -
mcp-research-routergithub.com/SpiritHerb/mcp-research-routerupdated this year
Intelligent MCP Research Router - Tool selection, search strategy planning, and parallel execution
npm MIT ★ 4 -
Focused MCP tools for Onchain Router text, image, and speech payments
npm MIT -
MCP-palvelin, joka täydentää LiteLLM:ää: mallin valinta tehtävätyypin mukaan, kulutuksen seuranta ja eval-vertailu tekoälyagenteille.
MIT -
oriora-router-mcpgithub.com/Orioralabs/oriora-router-mcpupdated this year
Oriora model-router as an MCP server — best AI model per task, executed on your own keys (BYOK)
MIT -
Local, cost-aware multi-model expert orchestration for Pi and Codex MCP hosts: discovers callable Pi models, dynamically assembles semantic expert teams from billing policy and capability profiles, executes in isolated worktrees, and returns structured results to the Main Agent
npm MIT -
MCP server: pick the cheapest Chinese AI model for a workload shape (coding, agent, video, image)
-
skill-router-g6github.com/ForYourHelpAI101/skill-router-g6updated this year
Self-learning skill router for AI agents. 6-stage pipeline with anti-hallucination protections. Model-agnostic MCP server.
MIT -
Project-scoped MCP memory router: per-project SQLite index and raw data, plus a global project registry.
MIT ★ 1 -
Open experience layer for conversational AI - routes user intent into safe, consented, stateful experiences. Claude MCP adapter included.
MIT -
search-routergithub.com/colaboy519/search-routerupdated this year
Smart search routing for AI agents — routes queries to the best provider based on language, query type, and provider capabilities
MIT -
Intelligent 3-tier task routing and active uncertainty fallback (Antiback) for Google Antigravity IDE. Powered by native Kev-0.8B (System One) and deterministic heuristics.
MIT ★ 1 -
Cost-aware, local-first skill routing for AI agents. MCP server + CLI + Claude Code hook. Tested on 141 real skills.
npm MIT ★ 1 -
MCP server that routes Claude tool calls to multiple Obsidian vaults (local or remote) over the Local REST API plugin.
Apache-2.0 ★ 1 -
claude-model-routergithub.com/korelex/claude-model-routerupdated this year
Routes each task to the cheapest Claude model that still clears the quality bar — a Claude Code skill + MCP server that optimizes total cost to an acceptable answer, not sticker price.
MIT ★ 3 -
context-routergithub.com/mohankrishnaalavala/context-routerupdated this year
Memory-aware context engine for AI coding agents — up to 91% fewer tokens, 17/18 rank-1 across 6 OSS projects. MCP-native, multi-repo, with persistent observations & decisions.
pypi NOASSERTION ★ 11