Predictive Maintenance MCP Server

Give any AI assistant the ability to analyze vibration data, detect machinery faults, and generate professional diagnostic reports — through natural conversation.

An open-source MCP server and predictive maintenance AI agent that turns LLMs into condition monitoring assistants. Engineers describe what they need in plain language; the AI calls the right analysis tools and delivers results — bearing fault detection, risk assessment, anomaly detection, and remaining useful life estimation. Also available as a Claude Code plugin with 8 diagnostic skills. It's designed to support and accelerate expert decision-making.

Who is this for?

• Reliability & maintenance engineers who want fast vibration diagnostics in plain language — no coding required. It augments and accelerates expert judgment; it doesn't replace it.
• Developers & industrial-AI practitioners who want to expose predictive-maintenance workflows as MCP tools and build on top of them.
• Researchers & students working on bearing fault diagnosis, condition monitoring, or MCP / agent tooling.

Quick Start

Get running in ~3 minutes. On Windows, one script wires everything into Claude Desktop — it installs the venv, pre-compiles dependencies, and writes claude_desktop_config.json for you (OneDrive / cloud-sync paths included):

git clone https://github.com/LGDiMaggio/predictive-maintenance-mcp.git
cd predictive-maintenance-mcp
.\setup_claude.ps1

Restart Claude Desktop, then try:

"Load real_train/OuterRaceFault_1.csv and check if the bearing is healthy."

pip install predictive-maintenance-mcp

{
  "mcpServers": {
    "predictive-maintenance": {
      "command": "/full/path/to/uvx",
      "args": ["predictive-maintenance-mcp"],
      "env": { "UV_LINK_MODE": "copy" }
    }
  }
}

More options: install from source · VS Code setup · Docker / HTTPS deployment · use with local LLMs (Ollama)

See It in Action

Full diagnostic workflow: load signal → spectral analysis → fault detection → severity assessment → report generation

What Can It Do?

Upload a vibration signal → get a professional diagnosis through conversation.

The AI doesn't guess — it calls 37 specialized MCP endpoints (34 tools + 3 prompts) running locally on your machine. Every signal is referenced by a single signal_id handle from load to report. Your data never leaves your infrastructure.

Claude Code Plugin

The project includes a plugin for Claude Code with domain-specific skills that activate automatically during conversation. Install it and Claude gains guided diagnostic workflows, autonomous agents, and quick commands.

/plugin marketplace add LGDiMaggio/predictive-maintenance-mcp
/plugin install predictive-maintenance@predictive-maintenance-marketplace

Claude Code plugin: domain skills activate automatically, slash commands for quick diagnostics

Skills (8) — activate automatically based on context

Agents (2) — run autonomously for complex tasks

Commands (3) — quick entry points

Reports

All analysis tools generate interactive HTML reports you can open in any browser — pan, zoom, hover for details. Also supports structured Word (.docx) exports.

Sample Data Included

The project ships with 20 real bearing vibration signals from production machinery tests — ready to use out of the box.

• Training set: 2 healthy baselines + 12 fault signals (inner race, outer race)
• Test set: 1 healthy baseline + 5 fault signals

Try: "Load real_train/OuterRaceFault_1.csv and diagnose the bearing fault."

Full dataset documentation: data/README.md

Architecture

          YOU (natural language)
               │
               v
     LLM (Claude, GPT, Ollama...)
     understands intent, selects tools
               │
               v  ── Model Context Protocol ──
    ┌──────────────────────────────┐
    │    Predictive Maintenance    │
    │         MCP Server           │
    │                              │
    │  Signal Analysis    Reports  │
    │  Fault Detection    ML       │
    │  Severity Rating    RAG Docs │
    └──────────────────────────────┘
               │
               v
       YOUR DATA (stays local)
    signals · manuals · models

The codebase follows a modular architecture organized around the ISO 13374 Six-Block Diagnostic standard — signal acquisition, processing, diagnostics, prognostics, and decision support as separate sub-packages.

src/predictive_maintenance_mcp/
├── mcp_tools/                 # MCP endpoint registration (37 MCP endpoints)
│   ├── acquisition_tools.py   # Signal loading & management
│   ├── analysis_tools.py      # Spectral & statistical analysis
│   ├── diagnostics_tools.py   # Fault detection, ML, document search
│   ├── report_tools.py        # HTML/DOCX report generation
│   ├── prompts.py             # Guided diagnostic workflows
│   └── _utils.py              # Shared utilities
├── signal_acquisition/        # Multi-format loaders (CSV, MAT, WAV, NPY, Parquet)
├── signal_processing/         # Spectral analysis & feature extraction
├── diagnostics/               # Bearing/gear analysis, ISO standards
├── decision_support/          # Evidence-based diagnosis pipeline
├── prognostics/               # RUL estimation (linear, exponential, Kalman) & trend analysis
├── rag.py                     # Document indexing & search (FAISS/TF-IDF)
├── models.py                  # Pydantic data models
├── server.py                  # MCPServer entry point
└── config.py                  # Configuration management

Key design choices:

• Privacy-first — raw vibration data never leaves your machine; only computed results flow to the LLM
• LLM-agnostic — works with Claude, ChatGPT, Microsoft Copilot Studio, or any MCP-compatible client. Use Ollama for fully air-gapped deployments
• Modular — use only the tools you need, extend with your own

Documentation

Testing

86% test coverage across Windows, macOS, and Linux (Python 3.11 & 3.12).

pytest                                  # run all tests
pytest --cov=src --cov-report=html      # with coverage report

20+ test files covering signal analysis, fault detection, severity assessment, ML models, report generation, RAG search, and real bearing fault data validation.

Benchmark

A blind, reproducible diagnostic-accuracy benchmark on the public CWRU Bearing Data Center dataset (12 kHz drive-end subset: 60 fault records + 4 normal baselines). Fault labels never reach the system under test — signals enter under opaque ids, a separate scorer is the only label reader, and blindness, checksum integrity, and determinism are enforced by CI-run guard tests, not prose. Results are stratified by the per-record diagnosability grades of the Smith & Randall (2015) reference study, so records that study found undiagnosable by any classical method are reported separately instead of inflating or deflating the headline.

On records the reference study grades clearly diagnosable (Y1+Y2, 44 records): characteristic fault frequency detected on 44/44, correct fault ranked first on 34/44 (77.3%), and 9/9 on the textbook-signature (Y1) stratum. On the 4 healthy baselines, 2 records raised a false indication under the same criterion.

The numbers above are read from the committed, re-runnable artifact (results.json) and drift-guarded by CI: every value is bound to its key in the artifact, and a mismatch fails the build. Methodology, blind protocol, and honest-benchmarking notes: docs/benchmark-methodology.md. Reproduce with:

python -m benchmarks.cwru all

Roadmap

• 37 MCP endpoints (34 tools, 3 prompts) with modular architecture and a single signal_id handle
• Claude Code plugin (8 skills, 2 agents, 3 commands)
• 86% test coverage, CI/CD on 3 platforms
• Docker + SSE/HTTP transport for enterprise deployment
• Semantic document search (FAISS + TF-IDF)
• Blind, reproducible diagnostic benchmark on the CWRU dataset (extensible to Paderborn)
• Customizable severity thresholds
• Remaining useful life (RUL) estimation from repeated measurements (linear, exponential, Kalman)
• Trend analysis and degradation onset detection
• Multi-signal trending and historical comparison
• Real-time streaming (MQTT/Kafka)
• Fleet dashboard for multi-asset monitoring
• CMMS integration (SAP, Maximo, Infor)

Ideas? Open a discussion or create an issue.

Are you using this?

I'd genuinely love to know. Whether you ran it on real machinery or just tried the sample data, drop a line in Discussions — one sentence about your machine or use case is enough. Real-world feedback directly shapes what gets built next.

Related

claude-stwinbox-diagnostics — Extends this project by connecting a physical edge sensor (STEVAL-STWINBX1) to Claude via MCP, with Claude Skills for guided condition monitoring. Same analysis engine, real hardware, operator-friendly reports.

Contributing

Contributions welcome from everyone — not just programmers. Domain experts, technical writers, and testers are equally valued. See CONTRIBUTING.md for paths tailored to your background.

Quick start: browse Issues for good first issue or help wanted labels.

Citation

@software{dimaggio_predictive_maintenance_mcp_2025,
  title   = {Predictive Maintenance MCP Server},
  author  = {Di Maggio, Luigi Gianpio},
  year    = {2025},
  version = {0.12.0},
  url     = {https://github.com/LGDiMaggio/predictive-maintenance-mcp},
  doi     = {10.5281/zenodo.17611542}
}

License

MIT — see LICENSE. Sample data is CC BY-NC-SA 4.0 (non-commercial); for commercial use, replace with your own machinery data.

Acknowledgments

MCP Python SDK (descended from FastMCP) · Model Context Protocol by Anthropic · Sample data from MathWorks · Core development assisted by Claude

An open-source predictive maintenance AI agent and condition monitoring copilot — built to support reliability engineers and the developer community.