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-fine-tune-exportergithub.com/joeyen-atscale/mcp-fine-tune-exporterupdated this year
Filter the AtScale MCP trace store for high-quality MQO interactions, exported as OpenAI chat JSONL for fine-tuning
-
fine-tuning-osgithub.com/Casius999/fine-tuning-osupdated this year
Zero-Data fine-tuning operations MCP server - 64 tools for the full LLM fine-tuning delivery lifecycle, callable by Claude Code
Apache-2.0 -
MCP server to fine-tune LoRA, export GGUF, and deploy local LLMs on Onde Inference.
npm -
tuningengines-cligithub.com/cerebrixos-org/tuning-engines-cliupdated this year
Tuning Engines CLI, MCP server, and Python agent runtime adapters for governed model, agent, skill, and MCP workflows. Fine-tune open-source LLMs, run inference, manage datasets/evaluations, and connect LangGraph or Temporal while Tuning Engines handles p
npm MIT -
@mseep/unsloth-serverstale
MCP server for Unsloth - a library for optimizing and fine-tuning LLMs
npm Apache-2.0 -
QLoRA fine-tuned email assistant with a full evaluation suite: pairwise LLM-judge comparison, fact-preservation testing, generalization checks, and FastAPI/MCP serving.
★ 1 -
tuneforgegithub.com/JustVugg/tuneforgeupdated this year
An MCP server for dataset generation, fine-tuning, RL, and evaluation of LLMs — directly from your coding agent.
MIT ★ 17 -
LLM projects: RAG with citations over mine safety regulations, an agent with a custom MCP server, LoRA/DPO fine-tuning, and an LLM evaluation harness.
-
industrial-data-analyst-agentgithub.com/MohamedAdelAbdelrahman/industrial-data-analyst-agentactive
LLM agent for industrial sensor data: LangGraph, own MCP server, fine-tuned Text-to-SQL (QLoRA), RAG, sandboxed Python, eval
-
synqtune-mcpactive
MCP server for SynqTune: check runs and comparison reports, estimate costs and launch fine-tuning pipelines in your own AWS account from your coding agent.
pypi LicenseRef-SynqTune-Proprietary -
Fine-tuned NL-to-SQL specialist served as an MCP tool, benchmarked against prompting Claude directly on accuracy, latency, and cost
★ 1 -
wpa-poster-sdxlgithub.com/nishankpt/wpa-poster-sdxlupdated this year
SDXL LoRA fine-tuned on public-domain WPA posters + a CLIP/FID/KID eval harness + MCP server. Trained and evaluated end-to-end on an 8GB laptop GPU.
MIT -
Train-in-Silencegithub.com/hlpun/Train-in-Silenceupdated this year
The first Task-Aware MCP server and automated VRAM calculator for LLM fine-tuning. Instantly snipe the cheapest, fastest GPUs across 10+ cloud providers.
pypi MIT ★ 104 -
Open-source AI agent that qualifies & responds to inbound leads in seconds — multi-agent (LangGraph) + a LoRA-fine-tuned intent classifier + MCP server. Bring your own keys.
MIT ★ 1 -
@iflow-mcp/secfathy-offensivesetgithub.com/pentesterflow/offensivesetupdated this year
MCP server for generating high-quality offensive security datasets for LLM fine-tuning. Produces realistic multi-turn pentesting conversations with tool calls, reasoning chains, and professional reports.
npm MIT -
Turn any website into a quality-scored LLM fine-tuning dataset: crawl, chunk, generate Q&A, score, export to JSONL/Parquet/HF Hub. CLI plus MCP server.
MIT ★ 2 -
clinical-trial-matchergithub.com/anumsagheer01/clinical-trial-matcherupdated this year
MCP-powered clinical trial matching platform. Searches 92,746 trials using OpenSearch hybrid retrieval, fine-tuned entity extraction (PEFT/LoRA), TensorFlow query routing, and Claude tool calling. Dockerized with CI/CD and Kubernetes manifests.
★ 1 -
Continuous LoRA fine-tuning as a local service. Watches source files for changes, retrains on change, and hot-swaps the adapter into a live OpenAI-compatible inference endpoint. Pipeline control via MCP.
MIT ★ 3 -
Linter for LLM fine-tuning datasets: catches broken chats and tool calls, context overflow, leaked keys, PII and eval contamination before you train. CLI, Python API, MCP server, GitHub Action.
pypi NOASSERTION ★ 5 -
nixtla-mcpgithub.com/VaclavRut/nixtla-mcpupdated this year
MCP server for multitenant Nixtla TimeGPT forecasting with fine-tuning, rolling backtests, and anomaly detection. HTTP-based JSON-RPC 2.0 API with MongoDB tracking and Vercel deployment.
MIT -
LoRA fine-tuned Qwen3-1.7B for multi-hop QA: hybrid retrieval + reranking, iterative search, MCP server, OpenTelemetry. Matches Qwen3-8B on HotpotQA (EM 10.4 → 58.4) at ~8× the throughput.
MIT -
RAG + LLM system for first-pass German contract compliance screening: hybrid BGB/DSGVO retrieval (RRF, cross-encoder reranking), LoRA fine-tuned DistilBERT clause classifier, evidence sufficiency checks, structured Claude reasoning, citation verification, MCP, PDF reports.
MIT ★ 1 -
Local mirror of AI-dev-tool changelogs (Anthropic, Cursor, Aider, Continue, MCP SDKs, etc.) with hybrid FTS5 + sqlite-vec retrieval, MCP server, curated cross-product synergies, and a fine-tuned changelog→action synthesizer (cs-actions:v1).
npm MIT -
Multi-tenant agent for telecom billing disputes: classifies disputes, retrieves tariff and contract clauses alongside invoice and call records, and makes cited credit decisions, with deterministic credit calculation, fraud escalation and CI-gated evals. LangGraph, MCP, fine-tuned classifier vs LLM baseline
MIT -
dsh-unslothactive
Unsloth for DeepSeek Harness: local LLM fine-tuning and dataset generation. Bundles a zero-dependency Python MCP server and mounts DSH's own MCP client, exposing model inventory, Data Recipe dataset generation, LoRA/QLoRA training, GGUF export and inferen
npm MIT -
yojana-sahayak-assistantgithub.com/Subh24ai/yojana-sahayak-assistantupdated this year
Fully offline, multilingual voice AI agent that helps Indian citizens discover government welfare schemes in Hindi & English. Uses Whisper ASR, fine-tuned Qwen2.5 (QLoRA), FAISS RAG, and an MCP tool server — all designed for air-gapped deployment. Covers 2,872 schemes from myscheme.gov.in.
★ 1 -
CMM_Toolsgithub.com/Redliana/CMM_Toolsupdated this year
Critical Minerals & Materials (CMM) tools monorepo - MCP servers, data libraries, fine-tuning pipelines
MIT -
Gridline MCP server (@gridlinegpu/mcp): fine-tune, evaluate and batch open models from your agent
npm MIT -
unsloth-mcp-servergithub.com/OtotaO/unsloth-mcp-serverupdated this year
An MCP server for Unsloth - a library that makes LLM fine-tuning 2x faster with 80% less memory
★ 8 -
bias-mcp-servergithub.com/yourusername/bias-mcp-serverupdated this year
MCP server for political bias detection using fine-tuned T5 model with LoRA adapters
pypi MIT -
ml-mcpgithub.com/PushPullCommitPush/ml-mcpupdated this year
A comprehensive MCP (Model Context Protocol) server for ML model training, fine-tuning, and experimentation. Transform your AI assistant into a full ML engineering environment.
NOASSERTION ★ 4 -
BASTION-MCPgithub.com/bastionfintech-spec/BASTION-MCPupdated this year
Risk Intelligence MCP Server for Claude Agents | 49 tools, 560+ signals, fine-tuned 72B AI model
MIT ★ 1 -
AI que diseña en Photoshop con tu estilo, a partir de tus PSDs anteriores. Sin fine-tuning: captura tu estilo como capa de conocimiento editable.
-
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 -
DecompGTIgithub.com/yuvraj-rathod-1202/DecompGTIupdated this year
DecompGTI is a graph-reasoning framework that fine-tunes small language models to parse natural-language graph queries into structured JSON plans executed by deterministic algorithms on a FastMCP server.
MIT -
MCP server that lets AI assistants call the GoHumanize Open Humanizer, an open Qwen3-4B fine-tune that rewrites AI-styled English prose into more natural human writing.
npm Apache-2.0 ★ 1 -
JigsawStack vOCRno update data
OCR + AI = Magic: Extract data from any image in a consistent structure with fine-tuned vLLMs for the highest accuracy. Learn more about AI backed OCR here: https://jigsawstack.com/docs/api-reference/ai/vocr Get your jigsawstack API key here: https://jigsawstack.com/dashboard/
smithery -
Portable, privacy-first AI context layer & MCP server. Classifies personal data on-device via fine-tuned Gemma 3 & WebGPU, keeping private data offline.
AGPL-3.0 ★ 1 -
Zero-deploy, Git-native content management CLI for Markdown stories and knowledge bases. Built-in MCP Server for AI integration, EPUB export, and structure-aware data pipelines for LLM fine-tuning.
npm MIT -
openai-mcp-servergithub.com/salviz/openai-mcp-serverupdated this year
MCP server exposing the full OpenAI API (Responses, Chat, Images, Audio, Embed, Files, Batch, Vector Stores, Fine-tune, Uploads) — 42 tools for Claude Code, Cursor, and any MCP client.
MIT ★ 2 -
Convert AI coding CLI sessions between Codex, Claude Code, Pi, OpenCode, Devin, and Factory. SDK, CLI, MCP chat recall, and trace export for fine-tuning.
pypi MIT ★ 22 -
knowledge-reduce-coregithub.com/nelslindahlx/knowledge-reduce-coreupdated this year
High-performance Knowledge-RAG pipeline featuring self-correcting LLM critique loops, MLX fine-tuning, real-time watcher daemons, pluggable local KùzuDB/cloud Neo4j graph stores, and an interactive 3D WebGL dashboard.
NOASSERTION -
LoRA-FAQ Hamilton is a lightweight project that fine-tunes GPT-2 with LoRA on the Alexander Hamilton Wikipedia page, enabling it to return direct quotes based on user input. The model is exposed as an MCP server tool, making Hamilton quotes accessible to any MCP-compatible client.
-
Where's my hug at? Tell your AI agent what you want a model to do. It finds what Hugging Face already has for it, what to fine-tune, and what's just code, with every model checked live.
npm MIT -
MCP server for daily change log and tracking of daily task. Helps to create professional fine tuned change logs on every tasks on a daily basis to share it to the client.
-
PentestAIgithub.com/mhmdtaha091/PentestAIupdated this year
Autonomous web-pentest agent on consumer hardware — LangGraph + 23-tool MCP server + HippoRAG knowledge graph, driving a fine-tune-free Qwen2.5-Coder-14B on 12GB VRAM. Clears every Low/Medium module and 84.6% at High on DVWA.
NOASSERTION ★ 1 -
A production-grade, MCP-compliant agentic forecasting tool. Designed for Agentic AI orchestrators to autonomously resolve inventory shortages, evaluate demand risk, and execute quantile-bounded supply-chain decisions—without requiring local fine-tuning.
MIT -
Where my hug at? Tell your AI agent what you want a model to do. It finds what Hugging Face, Kaggle, Zenodo, ModelScope and Roboflow already have for it, models and data alike, what to fine-tune, and what's just code, with every pick checked live.
npm MIT -
Occupation classification and job-title normalization for the O*NET-SOC taxonomy. A fine-tuned retrieve-and-rerank model maps job titles, resumes, postings and skill lists to ranked SOC codes with calibrated confidence. 51% top-1 vs 21% TF-IDF. Python, CLI, REST API, MCP server. Runs locally.
pypi MIT ★ 2 -
Graph-Augmented Code Intelligence RAG engine pairing deterministic AST call graphs with dense vector search to boost direct-callee Recall@8 from 0.0% to 53.3% (+53.3 pp) under a 5.8ms p95 latency bound. Powered by a QLoRA fine-tuned Qwen2.5 adapter (70.02 CodeBLEU), quality verification harness, and production Model Context Protocol (MCP) server.
MIT -
Adaption plugin for Claude Code — datasets, training, and AutoScientist workflows
MIT -
AI that trains AI: analyse your data and hardware, pick a method, train, tune and deploy models on your own machine — from a CLI, MCP, a web GUI or one sentence.
pypi MIT -
jw-agent-toolkitgithub.com/elimorals/jw-agent-toolkitupdated this year
Toolkit agentic no oficial de JW: CLI, MCP, RAG, Skills, etc.
Apache-2.0 ★ 7 -
Train and call custom Zils decision models from your coding agent (Claude Code and Cursor plugin, MCP server)
MIT -
@mubit-ai/mcpactive
The MCP server for Mubit: manage tasks, calls, trained models, keys, settings and usage from your AI agent
npm Apache-2.0 -
Open, MCP-native training-data foundry: turn permitted documents into traceable, quality-gated SFT and preference datasets.
Apache-2.0 ★ 1 -
A local studio to tune and extend language models: versioned datasets, LoRA and QLoRA training on the allowed GPUs, merges, GGUF conversion, importance-matrix quantization, YaRN context extension and evaluation against the base model. MCP tools for assistants.
MIT -
Small decision models for your app's yes/no, choice and score questions: CLI, Python and TypeScript clients, MCP server
npm MIT -
Incident-triage agent that cites real evidence or abstains — model-agnostic, read-only MCP servers, frontier-competitive on a local 4B at $0.
pypi MIT ★ 1 -
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