Skip to content
Back to search

predictive-maintenance-marketplace

MCP serverGitHubIncomplete

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 16 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/LGDiMaggio/predictive-maintenance-mcp with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/LGDiMaggio/predictive-maintenance-mcp'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/lgdimaggio--predictive-maintenance-mcp/badge.svg)](/listings/lgdimaggio--predictive-maintenance-mcp)

HTML

<a href="/listings/lgdimaggio--predictive-maintenance-mcp"><img src="/api/listings/lgdimaggio--predictive-maintenance-mcp/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

One route, from a fresh clone to a first diagnosis in Claude Desktop. You need [Git](https://git-scm.com/), [uv](https://docs.astral.sh/uv/getting-started/installation/) (it fetches Python 3.11 by itself) and [Claude Desktop](https://claude.ai/download).

**1. Clone the repository.** The 20 sample signals come with the clone, not with the PyPI package or with the source ZIP and tarball archives generated from this revision: real bearing measurements from the MathWorks rolling-element bearing dataset, healthy baselines plus inner and outer race faults, each with its companion metadata ([data/README.md](data/README.md)).

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

**2. Install and connect.** On Windows, one script creates the environment, installs the server and writes the Claude Desktop entry (clones inside OneDrive or other synced folders are handled):

powershell -ExecutionPolicy Bypass -File setup_claude.ps1

<details> <summary><b>macOS</b></summary>

uv venv .venv --python 3.11
uv pip install --python .venv/bin/python .

Add this entry to `~/Library/Application Support/Claude/claude_desktop_config.json`, with the absolute path of your clone in place of `/path/to/predictive-maintenance-mcp`. If a `predictive-maintenance` entry is already there, replace it:

{
  "mcpServers": {
    "predictive-maintenance": {
      "command": "/path/to/predictive-maintenance-mcp/.venv/bin/python",
      "args": ["-m", "predictive_maintenance_mcp"],
      "env": { "PDM_PROJECT_DIR": "/path/to/predictive-maintenance-mcp" }
    }
  }
}

`PDM_PROJECT_DIR` tells the server where the sample signals and the reports live. Keep every path absolute: Claude Desktop launches servers with a minimal `PATH`.

</details>

**3. Restart and check.** Quit Claude Desktop completely (File, then Quit) and start it again. In a new chat ask:

> *"List the available vibration signals."*

The 20 included signals should appear. If they do not, see [INSTALL.md, Troubleshooting](INSTALL.md#troubleshooting).

**4. First diagnosis.** Try:

> *"Load real_train/OuterRaceFault_1.csv. The shaft runs at 1500 rpm: check the bearing against the fault frequencies declared in its metadata, assess the ISO severity and generate the envelope report."*

What comes back is the server's evidence, not the model's opinion: in the envelope spectrum the outer-race and inner-race lines both stand above the local noise floor, each with a further line of its own, so the evidence of this single spectrum is consistent with more than one fault type, and the check lists both with their levels without singling either out, while neither the ball nor the cage line stands above the floor; the ISO 20816-3 severity zone of the broadband velocity, with the machine group and support type the assessment assumed; and the path of an HTML report in `reports/` with those frequencies marked on the envelope spectrum. The call on the machine stays with you.

The file name carries the fault label, but the server never reads names. Ask which frequencies matched and with what deviation: the answer rests on the spectrum alone.

**Use your own data.** Put a single-column CSV without a header row where the server can read it (a folder of yours under `data/signals/`, or an absolute path), add a `<name>_metadata.json` next to it with `sampling_rate` and `signal_unit` (for example `"g"`), and ask for the analysis stating the shaft speed. Details and traps: [Use Your Own Data](docs/QUICKSTART_ENGINEER.md#use-your-own-data). An export with a header, a units row, a preamble or several channels goes through the signal import instead, which reads the file as it is and asks you for every value it cannot verify (on the main branch, not yet in a published release): [Signal Import Guide](docs/IMPORT_GUIDE.md).

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 14 lines
  1. {
  2. "name": "predictive-maintenance-marketplace",
  3. "owner": {
  4. "name": "Luigi Di Maggio"
  5. },
  6. "plugins": [
  7. {
  8. "name": "predictive-maintenance",
  9. "source": "./plugin",
  10. "description": "Complete predictive maintenance toolkit for vibration analysis, bearing/gear fault diagnosis, anomaly detection, and ISO standards compliance.",
  11. "version": "1.4.1"
  12. }
  13. ]
  14. }
README.md · README · 329 lines
  1. <p align="center">
  2. <picture>
  3. <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/LGDiMaggio/predictive-maintenance-mcp/main/assets/banner-dark.png">
  4. <img src="https://raw.githubusercontent.com/LGDiMaggio/predictive-maintenance-mcp/main/assets/banner-light.png" alt="Predictive Maintenance MCP — ask your machines what's wrong, with evidence, not guesses" width="100%">
  5. </picture>
  6. </p>
  7. # Predictive Maintenance MCP Server
  8. <!-- mcp-name: io.github.LGDiMaggio/predictive-maintenance-mcp -->
  9. [![Python 3.11+](https://img.shields.io/badge/python-3.11+-blue.svg)](https://www.python.org/downloads/)
  10. [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.17611542.svg)](https://doi.org/10.5281/zenodo.17611542)
  11. [![Tests](https://github.com/LGDiMaggio/predictive-maintenance-mcp/actions/workflows/tests.yml/badge.svg)](https://github.com/LGDiMaggio/predictive-maintenance-mcp/actions/workflows/tests.yml)
  12. [![codecov](https://codecov.io/gh/LGDiMaggio/predictive-maintenance-mcp/branch/main/graph/badge.svg)](https://codecov.io/gh/LGDiMaggio/predictive-maintenance-mcp)
  13. [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
  14. [![LGDiMaggio/predictive-maintenance-mcp MCP server](https://glama.ai/mcp/servers/LGDiMaggio/predictive-maintenance-mcp/badges/score.svg)](https://glama.ai/mcp/servers/LGDiMaggio/predictive-maintenance-mcp)
  15. **Give your AI assistant evidence-based vibration diagnostics — machinery fault detection, ISO-cited severity, and diagnostic reports built to support and accelerate expert decision-making.**
  16. An open-source [MCP](https://modelcontextprotocol.io/) server that turns LLMs into condition monitoring assistants for reliability engineers. Its core design rule: **the server refuses to guess**. No diagnosis is ever inferred from filenames or statistical parameters alone — a fault indication requires matching spectral evidence. Every severity claim cites ISO 20816-3, and the evaluative wording in reports is authored by the server, not improvised by the model. The AI orchestrates the analysis and presents the evidence — detected fault frequencies, matched fault patterns, severity zones — while the final judgment stays with the engineer. Also available as a **Claude Code plugin** with 9 diagnostic skills.
  17. ---
  18. ## Choose Your Path
  19. | You are | Start here |
  20. |---------|------------|
  21. | **Reliability / maintenance engineer** — diagnostics in plain language, no coding | [Quick Start](#quick-start) below · then [Engineer's guide](docs/QUICKSTART_ENGINEER.md) for your own data, manuals and workflows |
  22. | **AI / MCP developer** — run, integrate, and extend the server | [Developer's Quickstart](docs/QUICKSTART_DEVELOPER.md) · [Installation](INSTALL.md) |
  23. | **Researcher / evaluator** — how the numbers are measured | [Benchmark Methodology](docs/benchmark-methodology.md) · [Benchmark](#benchmark) below |
  24. ---
  25. ## What Can It Do?
  26. **Point the AI at a vibration signal → get the evidence behind the fault — detected frequencies, matched fault patterns, ISO-cited severity — to support your call.**
  27. | You say | The AI does |
  28. |---------|-------------|
  29. | *"What does the vibration say about this bearing?"* | Loads the signal, tests each characteristic fault line against the noise floor of the envelope spectrum, reports what the evidence is consistent with and what it cannot exclude, cites the ISO 20816-3 severity zone |
  30. | *"Generate a full diagnostic report"* | Produces an interactive HTML report with charts, fault markers, and server-authored severity wording |
  31. | *"Extract specs from test_pump_manual.pdf and diagnose the signal"* | Reads the equipment manual, looks up the bearing model, calculates expected fault frequencies, tests each of them in the signal's envelope spectrum |
  32. | *"Train an anomaly detector on my healthy baselines, then flag anomalies"* | Trains a model on your normal data, scores new signals, flags outliers for your review |
  33. | *"What changed on pump P-101 since the baseline?"* | Reads the asset's recorded history, compares the latest acquisitions with the declared reference, and reports each indicator's change with the criterion it applied |
  34. The AI doesn't guess: it calls **48 specialized MCP endpoints** (44 tools + 4 prompts) running locally on your machine. These counts describe the main branch; tools listed under Unreleased in the [CHANGELOG](CHANGELOG.md) are not yet in the published PyPI package. Every signal is referenced by a single `signal_id` handle from load to report. Your data never leaves your infrastructure.
  35. Full endpoint reference, grouped by category: **[Tool Catalog](docs/TOOL_CATALOG.md)**.
  36. ---
  37. ## Quick Start
  38. One route, from a fresh clone to a first diagnosis in Claude Desktop. You need [Git](https://git-scm.com/), [uv](https://docs.astral.sh/uv/getting-started/installation/) (it fetches Python 3.11 by itself) and [Claude Desktop](https://claude.ai/download).
  39. **1. Clone the repository.** The 20 sample signals come with the clone, not with the PyPI package or with the source ZIP and tarball archives generated from this revision: real bearing measurements from the MathWorks rolling-element bearing dataset, healthy baselines plus inner and outer race faults, each with its companion metadata ([data/README.md](data/README.md)).
  40. ```bash
  41. git clone https://github.com/LGDiMaggio/predictive-maintenance-mcp.git
  42. cd predictive-maintenance-mcp
  43. ```
  44. **2. Install and connect.** On Windows, one script creates the environment, installs the server and writes the Claude Desktop entry (clones inside OneDrive or other synced folders are handled):
  45. ```powershell
  46. powershell -ExecutionPolicy Bypass -File setup_claude.ps1
  47. ```
  48. <details>
  49. <summary><b>macOS</b></summary>
  50. ```bash
  51. uv venv .venv --python 3.11
  52. uv pip install --python .venv/bin/python .
  53. ```
  54. Add this entry to `~/Library/Application Support/Claude/claude_desktop_config.json`, with the absolute path of your clone in place of `/path/to/predictive-maintenance-mcp`. If a `predictive-maintenance` entry is already there, replace it:
  55. ```json
  56. {
  57. "mcpServers": {
  58. "predictive-maintenance": {
  59. "command": "/path/to/predictive-maintenance-mcp/.venv/bin/python",
  60. "args": ["-m", "predictive_maintenance_mcp"],
  61. "env": { "PDM_PROJECT_DIR": "/path/to/predictive-maintenance-mcp" }
  62. }
  63. }
  64. }
  65. ```
  66. `PDM_PROJECT_DIR` tells the server where the sample signals and the reports live. Keep every path absolute: Claude Desktop launches servers with a minimal `PATH`.
  67. </details>
  68. **3. Restart and check.** Quit Claude Desktop completely (File, then Quit) and start it again. In a new chat ask:
  69. > *"List the available vibration signals."*
  70. The 20 included signals should appear. If they do not, see [INSTALL.md, Troubleshooting](INSTALL.md#troubleshooting).
  71. **4. First diagnosis.** Try:
  72. > *"Load real_train/OuterRaceFault_1.csv. The shaft runs at 1500 rpm: check the bearing against the fault frequencies declared in its metadata, assess the ISO severity and generate the envelope report."*
  73. What comes back is the server's evidence, not the model's opinion: in the envelope spectrum the outer-race and inner-race lines both stand above the local noise floor, each with a further line of its own, so the evidence of this single spectrum is consistent with more than one fault type, and the check lists both with their levels without singling either out, while neither the ball nor the cage line stands above the floor; the ISO 20816-3 severity zone of the broadband velocity, with the machine group and support type the assessment assumed; and the path of an HTML report in `reports/` with those frequencies marked on the envelope spectrum. The call on the machine stays with you.
  74. The file name carries the fault label, but the server never reads names. Ask which frequencies matched and with what deviation: the answer rests on the spectrum alone.
  75. **Use your own data.** Put a single-column CSV without a header row where the server can read it (a folder of yours under `data/signals/`, or an absolute path), add a `<name>_metadata.json` next to it with `sampling_rate` and `signal_unit` (for example `"g"`), and ask for the analysis stating the shaft speed. Details and traps: [Use Your Own Data](docs/QUICKSTART_ENGINEER.md#use-your-own-data). An export with a header, a units row, a preamble or several channels goes through the signal import instead, which reads the file as it is and asks you for every value it cannot verify (on the main branch, not yet in a published release): [Signal Import Guide](docs/IMPORT_GUIDE.md).
  76. > **Other clients**: Claude Code, VS Code, Cursor and any other MCP client: [INSTALL.md, Manual Setup](INSTALL.md#manual-setup-windows--macos--linux).
  77. ---
  78. ## Benchmark
  79. A blind, reproducible benchmark of the bearing check on the public
  80. [CWRU Bearing Data Center](https://engineering.case.edu/bearingdatacenter)
  81. dataset (12 kHz drive-end subset: 60 fault records + 4 normal baselines).
  82. Fault labels never reach the system under test — signals enter under opaque
  83. ids, a separate scorer is the only label reader, and blindness, checksum
  84. integrity, and determinism are enforced by CI-run guard tests, not prose.
  85. Results are stratified by the per-record diagnosability grades of the
  86. Smith & Randall (2015) reference study, so records that study found
  87. undiagnosable by any classical method are reported separately instead of
  88. inflating or deflating the headline.
  89. <!-- cwru-benchmark:start -->
  90. On the records the reference study grades clearly diagnosable (Y1+Y2, <!-- slot: headline.n_records -->44<!-- /slot --> records), the line tested for the labelled fault stood above the local noise floor on <!-- slot: headline.frequency_detection.hits -->44<!-- /slot -->/<!-- slot: headline.frequency_detection.total -->44<!-- /slot -->, and the labelled fault was supported, with a further line of its own, on <!-- slot: headline.label_supported.count -->44<!-- /slot -->/<!-- slot: headline.label_supported.total -->44<!-- /slot -->.
  91. Another fault type was supported as well on <!-- slot: headline.wrong_type_support.records -->44<!-- /slot -->/<!-- slot: headline.wrong_type_support.total -->44<!-- /slot -->; the check listed several types, the labelled one among them, on <!-- slot: headline.outcomes.several_types_with_label -->44<!-- /slot -->/<!-- slot: headline.outcomes.total -->44<!-- /slot --> and singled out no fault type on <!-- slot: headline.named_fault.none -->44<!-- /slot -->/<!-- slot: headline.named_fault.total -->44<!-- /slot -->.
  92. The labelled fault was the first candidate listed, by level over the noise floor, on <!-- slot: headline.strongest_candidate.correct -->40<!-- /slot -->/<!-- slot: headline.strongest_candidate.total -->44<!-- /slot --> (<!-- slot: strata.Y1.strongest_candidate.correct -->9<!-- /slot -->/<!-- slot: strata.Y1.strongest_candidate.total -->9<!-- /slot --> on the textbook-signature Y1 stratum); that order is not a ranking of likelihood, and the check names no fault from it.
  93. Of the <!-- slot: strata.ungraded.normals.total -->4<!-- /slot --> normal baselines, <!-- slot: strata.ungraded.normals.no_indication -->1<!-- /slot --> gave no indication and <!-- slot: strata.ungraded.normals.bearing_alert -->3<!-- /slot --> listed bearing candidates, <!-- slot: strata.ungraded.normals.named_fault -->1<!-- /slot --> of them singling out one fault type: counts on these recordings, not a specificity rate.
  94. <!-- cwru-benchmark:end -->
  95. The numbers above are read from the committed, re-runnable artifact
  96. ([results.json](benchmarks/cwru/results/results.json)) and drift-guarded by CI:
  97. every value is bound to its key in the artifact, and a mismatch fails the build.
  98. Methodology, blind protocol, and honest-benchmarking notes:
  99. [docs/benchmark-methodology.md](docs/benchmark-methodology.md). Reproduce with:
  100. ```bash
  101. python -m benchmarks.cwru all
  102. ```
  103. ---
  104. ## Claude Code Plugin
  105. The project includes a **plugin for Claude Code** with domain-specific skills that activate automatically during conversation.
  106. ```shell
  107. /plugin marketplace add LGDiMaggio/predictive-maintenance-mcp
  108. /plugin install predictive-maintenance@predictive-maintenance-marketplace
  109. ```
  110. <p align="center">
  111. <img src="assets/plugin.gif" alt="Animated demo: adding the predictive-maintenance marketplace, installing the plugin and opening its page, which lists its skills and agents" width="720">
  112. </p>
  113. The plugin adds **9 skills** that activate automatically based on context (bearing-diagnosis, gear-diagnosis, quick-screening, report-generation, anomaly-detection, signal-management, documentation-search, prognostics, experimental shaft-order-evidence), **2 agents** that run multi-step diagnostic workflows end-to-end and hand you the evidence (diagnostic-pipeline, signal-explorer), and **3 commands** for quick entry points (`/pm-diagnose`, `/pm-screen`, `/pm-report`).
  114. Full skill, agent, and command reference: [Plugin README](plugin/README.md).
  115. ---
  116. ## Reports
  117. All analysis tools generate **interactive HTML reports** you can open in any browser — pan, zoom, hover for details. Also supports structured Word (.docx) exports.
  118. <details>
  119. <summary><b>Report examples</b></summary>
  120. ![Example envelope analysis report, generated from a synthetic test signal](assets/envelope_analysis.png)
  121. ![Example ISO 20816-3 severity report, generated from a synthetic test signal](assets/iso.png)
  122. *Both examples were generated from a synthetic test signal with the `generate_test_signal` tool. They show what the reports contain; the Zone A result does not describe a real machine.*
  123. | Report Type | What it shows |
  124. |-------------|---------------|
  125. | Frequency spectrum | Peak detection, harmonic markers |
  126. | Envelope analysis | Bearing fault frequency matching |
  127. | Severity assessment | Vibration health zones (ISO 20816-3) |
  128. | Word document | Full diagnostic narrative with embedded charts |
  129. | PCA visualization | Multi-signal anomaly clustering |
  130. | Feature comparison | Side-by-side signal feature analysis |
  131. </details>
  132. ---
  133. ## Asset Health Ledger
  134. A single diagnosis answers "what does this signal show". The asset health ledger answers "what changed on this machine since the reference": every measurement loaded with a declared identity (asset, measurement point, acquisition instant, operating conditions) is recorded in a local append-only ledger with its derived indicators, compared with the healthy baseline the engineer declares, and judged against the reference band with the rule it applied stated in the result. Four tools carry the workflow: `declare_measurement_point`, `declare_healthy_baseline`, `get_asset_history` and `assess_asset_change`. Contract and operations: [Adapter Guide](docs/ADAPTER_GUIDE.md#declaring-a-measurement) · worked flow: [Examples](EXAMPLES.md#example-8-asset-history) · reference adapter: [STWIN.box](examples/adapters/stwinbox/README.md).
  135. ---
  136. ## Architecture
  137. ```
  138. YOU (natural language)
  139. │
  140. v
  141. LLM (Claude, GPT, Ollama...)
  142. understands intent, selects tools
  143. │
  144. v ── Model Context Protocol ──
  145. ┌──────────────────────────────┐
  146. │ Predictive Maintenance │
  147. │ MCP Server │
  148. │ │
  149. │ Signal Analysis Reports │
  150. │ Fault Detection ML │
  151. │ Severity Rating RAG Docs │
  152. └──────────────────────────────┘
  153. │
  154. v
  155. YOUR DATA (stays local)
  156. signals · manuals · models
  157. ```
  158. 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. Standards implemented: ISO 13374, ISO 20816-3, MIMOSA OSA-CBM. Module-level detail: [Architecture guide](docs/architecture.md).
  159. **Key design choices:**
  160. - **Privacy-first** — raw vibration data never leaves your machine; only computed results flow to the LLM
  161. - **LLM-agnostic** — works with Claude, ChatGPT, Microsoft Copilot Studio, or any MCP-compatible client. Use [Ollama](docs/OLLAMA_GUIDE.md) for fully air-gapped deployments
  162. - **Modular** — use only the tools you need, extend with your own
  163. ---
  164. ## Documentation
  165. | Guide | For |
  166. |-------|-----|
  167. | [Engineer's guide](docs/QUICKSTART_ENGINEER.md) | After the first diagnosis: your own data, machine manuals, workflows |
  168. | [Quickstart for Developers](docs/QUICKSTART_DEVELOPER.md) | Understand MCP, extend the server |
  169. | [Tool Catalog](docs/TOOL_CATALOG.md) | Every MCP endpoint, grouped by category |
  170. | [Adapter Guide](docs/ADAPTER_GUIDE.md) | Bring vendor/DAQ raw data in via explicit declarations |
  171. | [Plugin README](plugin/README.md) | Claude Code plugin installation and usage |
  172. | [HTTPS Deployment](docs/DEPLOYMENT.md) | Docker + HTTPS for enterprise environments |
  173. | [Ollama Guide](docs/OLLAMA_GUIDE.md) | Use with local LLMs (fully air-gapped) |
  174. | [Architecture](docs/architecture.md) | ISO 13374 block mapping and module design |
  175. | [Benchmark Methodology](docs/benchmark-methodology.md) | How the CWRU diagnostic benchmark is measured |
  176. | [Examples](EXAMPLES.md) | Complete diagnostic workflows |
  177. | [Installation](INSTALL.md) | Detailed setup and troubleshooting |
  178. | [Contributing](CONTRIBUTING.md) | How to contribute (all skill levels welcome) |
  179. | [Changelog](CHANGELOG.md) | Version history |
  180. ---
  181. ## Testing
  182. **85%+ test coverage**, enforced as a CI minimum, across Windows, macOS, and Linux (Python 3.11 & 3.12) — the current measured figure is on the codecov badge above.
  183. ```bash
  184. pytest # run all tests
  185. pytest --cov=src --cov-report=html # with coverage report
  186. ```
  187. 20+ test files covering signal analysis, fault detection, severity assessment, ML models, report generation, RAG search, and real bearing fault data validation.
  188. ---
  189. ## Roadmap
  190. - [x] 48 MCP endpoints on the main branch (44 tools, 4 prompts) with modular architecture and a single `signal_id` handle
  191. - [x] Claude Code plugin (9 skills, 2 agents, 3 commands)
  192. - [x] 85%+ test coverage enforced in CI, CI/CD on 3 platforms
  193. - [x] Docker + SSE/HTTP transport for enterprise deployment
  194. - [x] Semantic document search (FAISS + TF-IDF)
  195. - [x] Blind, reproducible diagnostic benchmark on the CWRU dataset (extensible to Paderborn)
  196. - [ ] Customizable severity thresholds
  197. - [x] Remaining useful life (RUL) estimation from repeated measurements (linear, exponential, Kalman)
  198. - [x] Trend analysis and degradation onset detection
  199. - [x] Asset health ledger: per-asset measurement history with a declared reference and qualified comparability
  200. - [ ] Multi-signal trending and historical comparison
  201. - [ ] Real-time streaming (MQTT/Kafka)
  202. - [ ] Fleet dashboard for multi-asset monitoring
  203. - [ ] CMMS integration (SAP, Maximo, Infor)
  204. > Ideas? [Open a discussion](https://github.com/LGDiMaggio/predictive-maintenance-mcp/discussions) or [create an issue](https://github.com/LGDiMaggio/predictive-maintenance-mcp/issues/new/choose).
  205. ---
  206. ## Are you using this?
  207. I'd genuinely love to know. Whether you ran it on real machinery or just tried the sample data, drop a line in [Discussions](https://github.com/LGDiMaggio/predictive-maintenance-mcp/discussions) — one sentence about your machine or use case is enough. Real-world feedback directly shapes what gets built next.
  208. ---
  209. ## Related
  210. **[claude-stwinbox-diagnostics](https://github.com/LGDiMaggio/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.
  211. ---
  212. ## Contributing
  213. Contributions welcome from **everyone** — not just programmers. Domain experts, technical writers, and testers are equally valued. See [CONTRIBUTING.md](CONTRIBUTING.md) for paths tailored to your background.
  214. **Quick start**: browse [Issues](https://github.com/LGDiMaggio/predictive-maintenance-mcp/issues) for `good first issue` or `help wanted` labels.
  215. ---
  216. ## Citation
  217. ```bibtex
  218. @software{dimaggio_predictive_maintenance_mcp_2025,
  219. title = {Predictive Maintenance MCP Server},
  220. author = {Di Maggio, Luigi Gianpio},
  221. year = {2025},
  222. version = {0.13.0},
  223. url = {https://github.com/LGDiMaggio/predictive-maintenance-mcp},
  224. doi = {10.5281/zenodo.17611542}
  225. }
  226. ```
  227. ## License
  228. MIT, see [LICENSE](LICENSE). The sample data in `data/` is licensed separately under CC BY-NC-SA 4.0 (non-commercial); its attribution and terms are in [data/LICENSE](data/LICENSE). For commercial use, replace it with your own machinery data.
  229. ## Acknowledgments
  230. [MCP Python SDK](https://github.com/modelcontextprotocol/python-sdk) (descended from [FastMCP](https://github.com/jlowin/fastmcp)) · [Model Context Protocol](https://modelcontextprotocol.io/) by Anthropic · Sample data by Eric Bechhoefer, redistributed by [MathWorks](https://github.com/mathworks/RollingElementBearingFaultDiagnosis-Data) · Core development assisted by [Claude](https://claude.ai)
  231. ---
  232. *An open-source predictive maintenance AI agent and condition monitoring copilot — built to support reliability engineers and the developer community.*

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.