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.
-
Agent Skills and an MCP server for GPU performance profiling, benchmarking, optimization, and reporting, with an inference focus.
MIT ★ 4 -
@ebowwa/mcp-modalupdated this year
Modal GPU Cloud MCP server - manage GPU workloads, training, and robot inference
npm MIT -
CPU-only LLM inference engine in pure Rust — 4-bit quantized models, hand-written AVX2 kernels, speculative decoding, and a local HTTP/MCP server. No GPU, no Python runtime.
cargo Apache-2.0 ★ 16 -
ATLASgithub.com/web3guru888/ATLASupdated this year
Active-inference Training with Learned Adaptive Stigmergy — Pure Rust AGI framework, 21 crates, 565 tests, 28 MCP tools, BF16 GPU inference (15.4 tok/s OLMo-3-7B on A100), OpenAI-compatible API, ZK proofs. Zero external dependencies.
Apache-2.0 ★ 3 -
mollagithub.com/tamnd/mollaactive
Local inference server in pure Mojo. Speaks OpenAI, Anthropic, and MCP. Models are OCI artifacts. Same source runs on CPU, NVIDIA, AMD, and Apple GPUs. Apache-2.0 all the way down.
Apache-2.0 -
Claude Code and Codex plugin for deeplit: manage GPU inference instances, volumes, models and billing from your coding agent, authorized by a scoped API key.
Apache-2.0 ★ 2 -
terradev-mcpgithub.com/theoddden/terradev-mcpupdated this year
Complete Agentic GPU Infrastructure for Claude Code — 192 MCP tools: Full training lifecycle, inference deployment with cost guardrails, Ray cluster management (Wide-EP, disaggregated P/D, parallelism strategy), vLLM lifecycle (start/stop/sleep/wake/infer
npm MIT -
runpod-cowork-plugingithub.com/Angelus174/runpod-cowork-pluginupdated this year
RunPod MCP plugin for Claude and Cowork. Manage GPU pods, Jupyter environments, AI jobs, and cloud inference workloads through natural language.
MIT -
Governed GPU inference ops (vLLM + Ray Serve): latency RCA, scaling, drain, 30 MCP tools (preview)
pypi MIT -
ppb-mcpgithub.com/paulplee/ppb-mcpupdated this year
Poor Paul's Benchmark as an MCP server. Queryable GPU inference data — quantization, throughput, VRAM, concurrent users — for Claude Desktop, Cursor, Windsurf, Cline, and any MCP client. Self-host or use the free hosted endpoint.
MIT -
Make your model serve faster on your GPU, and prove it. Skills for Claude Code, Codex and OpenCode, plus a CLI and MCP server: your agent picks and tunes inference engines, can write GPU kernels, and backs every number with client-timed benchmarks, anti-cheat gates and quality checks.
pypi Apache-2.0 -
Badgr — GPU Compute, AI Model Deployment & Agent Executionno update data
**Run GPU workloads, deploy AI models, access inference APIs, and execute AI agents using Badgr's MCP server.** Badgr is an AI compute and workload execution platform for developers, coding agents, and AI assistants. Run workloads on cloud GPUs or compatible customer-owned GPUs (BYO GPU), with cost controls, execution logs, and verified results. **What can Badgr do?** - **GPU Compute / AI Workloads:** Run Python scripts, machine learning jobs, batch inference, and supported ComfyUI workflows. - **AI Model Deployment:** Deploy and serve LLMs using compatible runtimes such as vLLM and SGLang. Get a verified, OpenAI-compatible inference endpoint. - **AI API / LLM Inference:** Make AI API requests with response validation, supported fallback routing, and bounded execution. - **AI Agent Execution:** Launch coding and testing agents, run tasks, check acceptance criteria, and retrieve execution receipts. **Key features:** Cloud GPU execution, BYO GPU support, model serving, AI agent jobs, workload verification, execution logs, cost limits, and deployment management. Connect Badgr to your MCP-compatible AI assistant to execute real workloads instead of just generating instructions. Get started: https://aibadgr.com **Requires a Badgr API key. GPU workloads and other paid execution may incur usage charges.**
smithery -
Agentic toolkit for building and optimizing high-performance inference engines tailored to your GPU and desired LLM.
pypi Apache-2.0 ★ 48 -
@vglu/tele-gpu-pilot-mcpupdated this year
MCP server exposing tele-gpu-pilot pod inference (chat / image / status) to Claude Code, Cursor, and any MCP-compatible client via stdio. Self-hosted Flux schnell + Gemma 4 + Qwen 2.5-Coder on RunPod A40.
npm MIT -
vultr-blade-mcpgithub.com/groupthink-dev/vultr-blade-mcpupdated this year
Vultr GPU compute infrastructure MCP server — VMs, bare metal, serverless inference, DNS, firewall, snapshots, billing with token-efficient defaults
npm MIT -
pulse-mcpgithub.com/pulsebenchmarks/pulse-mcpupdated this year
Open compute-pricing MCP server — daily GPU and inference-token medians, CC-BY 4.0.
Apache-2.0 -
parallelix-mcpgithub.com/parallelixnetwork/parallelix-mcpupdated this year
MCP server for the ParalleliX Compute API. Offload parallel inference sub-tasks to a distributed GPU network, paid in $PRLX credits.
npm MIT -
Command-line interface and Python SDK for VoltageGPU: confidential GPUs on demand (Intel TDX, tenant-side NVIDIA attestation) and an OpenAI-compatible inference API, billed per second.
MIT ★ 1 -
Profile llama.cpp-family GPU inference, run controlled experiments, and compare results with evidence labels. CLI, MCP server and Agent Skill.
MIT ★ 1 -
Semantic search for your vault and code, through the CLI or MCP. Inference runs on your GPU.
pypi MIT ★ 2 -
Runtime evidence that helps agents trace, profile, and burn down hotspots in application and native code, GPU kernels, and inference stacks.
pypi MIT ★ 243 -
MyAino update data
Decentralized AI inference on Base: OpenAI-compatible chat via community GPUs, paid in MYAI
smithery -
MCP server and private mode for coding agents on OffRouter private inference. Every model runs in a GPU TEE, and your prompts are encrypted to the enclave before they leave your machine.
npm MIT -
@gpuflow/mcpactive
GPU Flow MCP server — exposes European LLM inference as tools for agents.
npm MIT -
Local-first AI memory system for an Obsidian vault: cited LLM wiki, MCP agent, MoE inference on a 16 GB GPU
MIT -
dcp-mcpgithub.com/dhnpmp-tech/dcp-mcpupdated this year
DCP MCP server — run sovereign in-Kingdom inference, rent GPUs, and manage storage on DCP (Saudi Arabia's sovereign AI compute platform) from any MCP agent. Thin client over https://api.dcp.sa.
MIT -
hardware-probegithub.com/yamaru-eu/hardware-probeupdated this year
Expert hardware probe for AI, Gaming, and HPC. Analyzes GPU, VRAM, and Memory Bandwidth to optimize local LLM inference (Ollama) and provide actionable hardware upgrade recommendations.
npm Apache-2.0 ★ 7 -
learn-mcpgithub.com/fvegiard/learn-mcpupdated this year
AI Engineering Learning Reference MCP Server — 550+ searchable references for MCP, RAG, agents, fine-tuning, MLOps, prompt engineering, inference, vector DBs, GPU compute, autonomous agents, and Claude Code workflows
MIT -
k402-mcpgithub.com/Kali123411/k402-mcpupdated this year
MCP server for k402 agent-payable APIs on Kaspa — text tools, chain data, GPU inference, covenants-as-a-service, and RISC Zero zk attestation, paid per call in KAS/BTC/LTC/DOGE/BCH/DASH/ETH/USDC — no account or API key
pypi MIT ★ 1 -
MCP server for Stable Diffusion WebUI Forge Neo — generate images on your own GPU from any MCP-capable agent. Reads the loaded checkpoint, infers sampling parameters and prompt style from your settings and past generations, and writes the prompt.
MIT ★ 1 -
salad-mcp-servergithub.com/ethanke/salad_mcpupdated this year
MCP Server for SaladCloud GPU provider services
npm MIT -
gpusmarketactive
Rent real GPUs from the terminal. Agent-first CLI + built-in MCP server for GPUsMarket — signup, rent, chat, pool. Per-second billing, $0.50 minimum per rental.
npm SEE LICENSE IN LICENSE -
@gpusmarket/cliactive
Rent real GPUs from the terminal. Agent-first CLI + built-in MCP server for GPUsMarket — signup, rent, chat, pool. Per-second billing, $0.50 minimum per rental.
npm SEE LICENSE IN LICENSE -
Model Context Protocol server for AI VRAM calculation and GPU recommendation
npm MIT -
NVIDIA Triton Inference Server control plane — model repository manipulation, VRAM monitoring, and inference over gRPC via FastMCP 3.2
MIT ★ 1 -
Public metadata for the ComputeSage StackBench remote MCP server.
GPL-3.0 -
gpubrokergithub.com/wudaming00/gpubrokerupdated this year
Lightweight GPU resource broker for multi-project single-host ML workloads. HTTP API + Python client + MCP server for AI agents.
NOASSERTION -
genie-lockergithub.com/nhantour/genie-lockerupdated this year
Read-only MCP connector and public API for quality-gated, ephemeral private AI inference.
oci MIT -
Free experimental GPU tools for AI agents via MCP: image generation, audio transcription and text. Public clients, examples and integration guides.
★ 1 -
Serve, size and benchmark open LLMs on one 16 GB GPU: vLLM vs llama.cpp vs Ollama, KV cache and offload planning, MCP server
MIT -
MCP server for Gate402: pay-per-call agent APIs over x402. Open-model LLM inference, Base trading data, SEC filings and web scraping. 20 tools, free tier, no signup.
npm MIT ★ 1 -
capix-mcpgithub.com/capix-protocol/capix-mcpupdated this year
Capix MCP Server — deploy and manage private LLM instances, compute, websites, and verified workloads on the Capix network from any AI coding agent that supports MCP.
npm Apache-2.0 -
AI/ML and HPC at planetary scale. One API. Every Accelerator. Any Region. MIT-0 licensed.
MIT-0 ★ 55 -
ownrig-mcpgithub.com/ownrig/ownrig-mcpupdated this year
OwnRig MCP Server — AI hardware compatibility data for Claude, ChatGPT, Cursor, and any MCP-compatible assistant. 50 models, 25 devices, 9 machines, 663 compatibility entries.
npm MIT -
llm_mcpgithub.com/AuraFriday/llm_mcpupdated this year
MCP server that runs local LLMs (with full access to MCP tools included). Callable by Python to chain MCP tools with local intelligence.
Apache-2.0 ★ 3 -
FastMCP server for the Animagine XL 4.0 image generation experience, providing prompt validation, optimization, explanation, img2img, and checkpoint/LoRA management tools. Works for any Stable Diffusion XL based model.
MIT ★ 1 -
angruvadalgithub.com/anna-claudette/angruvadalupdated this year
RAM-Backed MCP Memory Architecture for Consumer LLM Inference — 900K token context on 16GB VRAM
★ 3 -
Peridot v1.6.0 [STABLE] | Sovereign AI Kernel. Offline, GPU accelerated Qwen framework: a 27B model fully on an 8GB GPU at 23.6 tok/s. Sandboxed skills & plugins, model tool use, opt in web search, MCP bridge for private data jobs from cloud agents, ZAT-SCS preemption, RAG. Zero data exfiltration. Engineered by uncoalesced.
MIT ★ 8 -
llama.cpp's llama-server + Web UI, compiled natively for your exact CPU/GPU in Docker, with a hardware guard and private local web search (SearXNG + MCP).
MIT ★ 2 -
Local AI desktop app: a Chromium browser with a GPU-aware llama.cpp runtime, OpenAI-compatible API, AI chat and agents, Stable Diffusion image/video, TTS and music, plus an MCP server so Claude Code, Codex and Cline can drive your browser. Windows, macOS, Linux.
NOASSERTION ★ 151