v0.7.1 — SSE/HTTP Transport for Enterprise HTTPS Deployment

What's New

Enterprise HTTPS deployment — The server now supports SSE and Streamable-HTTP transports, enabling Microsoft Copilot Studio and other networked MCP clients to connect via HTTPS.

Added

• SSE & Streamable-HTTP transport — --transport sse|streamable-http CLI flag (or MCP_TRANSPORT env var) with configurable --host / --port
• Docker Compose + Caddy — One-command deployment with automatic Let's Encrypt HTTPS certificates
• Deployment guide — New docs/DEPLOYMENT.md covering Docker, Caddy, nginx, Azure, and Copilot Studio integration
• sse optional dependency — pip install predictive-maintenance-mcp[sse] installs uvicorn

Changed

• Dockerfile rebuilt for SSE default (uvicorn, EXPOSE 8000)
• README updated with enterprise features and deployment documentation

Quick Start (SSE mode)

# Local
predictive-maintenance-mcp --transport sse --host 0.0.0.0 --port 8080

# Docker
docker compose up -d

Full Changelog: v0.7.0...v0.7.1

v0.7.0 — FAISS Vector Search, OCR & DOCX Reports

What's New

FAISS Vector Search

• search_documentation now uses FAISS + sentence-transformers for true semantic retrieval when installed
• Falls back to TF-IDF keyword search automatically when not installed
• Install: pip install predictive-maintenance-mcp[vector-search]

OCR for Scanned PDFs

• extract_text_from_pdf() automatically falls back to Tesseract OCR for pages with empty/minimal text layers
• Install: pip install predictive-maintenance-mcp[ocr] + Poppler on PATH

DOCX Diagnostic Reports

• New generate_diagnostic_report_docx MCP tool generates structured Word documents
• Includes statistics tables, FFT/envelope peaks, bearing frequencies, ISO evaluation, diagnostic summary
• Install: pip install predictive-maintenance-mcp[docx]

Other Changes

• 3 new optional dependency groups: vector-search, ocr, docx
• full extra now includes all optional features
• 27 MCP tools (was 26)
• Updated all documentation, CHANGELOG, README, GitHub Pages, INSTALL guide

Install / Upgrade

\\�ash
pip install --upgrade predictive-maintenance-mcp

Or with all optional features:

pip install --upgrade predictive-maintenance-mcp[full]
\\

Full Changelog: v0.6.0...v0.7.0

v0.5.0: Multi-format support, ML dedup, code quality & docs update

What's New in v0.5.0

Added

• Multi-format signal loading (CSV, MAT, WAV, NPY, Parquet) via unified load_signal_data()
• python -m predictive_maintenance_mcp entry point (__main__.py)
• Ollama integration guide (docs/OLLAMA_GUIDE.md)
• 3 Copilot Skills (bearing-diagnosis, quick-screening, report-generation)
• MANIFEST.in updated for new data formats

Changed

• ML training code deduplicated (4 shared helpers, ~163 statements removed)
• PyPDF2 migrated to pypdf
• ISO metadata double-read eliminated
• pd.read_csv calls hardened with engine/dtype/on_bad_lines
• pytest config consolidated into pyproject.toml (pytest.ini removed)
• Version bumped 0.4.1 → 0.5.0 across all files
• Comprehensive docs update (README, CHANGELOG, QUICKSTART, CONTRIBUTING, INSTALL)

Fixed

• Envelope analysis band_start/band_end validation
• FFT frequency axis off-by-one
• ISO unit-confirmation flow
• ML predict tool missing model guard
• Anomaly plot data alignment

Tests

• 44 passed, 3 skipped, 0 warnings

---

Install: pip install predictive-maintenance-mcp==0.5.0

v0.4.1 — Ecosystem Restructuring & MCP Registry

What's New in v0.4.x

🎯 Ecosystem Restructuring (v0.4.0)

• Persona-based documentation: Separate quickstart guides for Maintenance Engineers (zero-code) and AI Developers (architecture + tutorial)
• "Our Mission" & Architecture diagram in README — explains the "why" of MCP for industrial diagnostics
• Revamped CONTRIBUTING.md with 4 contribution paths: Domain Expert, Developer, Writer, Tester
• GitHub Issue/PR templates: Bug reports, feature requests, domain validation (no-code), good first issues
• Actionable Roadmap with links to Issues/Discussions
• MCP Registry support: Automated publishing to the official MCP server registry

🔧 Fixes (v0.4.1)

• Aligned all version strings after merge with v0.3.3–v0.3.4 changes
• Fixed server.json description length for MCP Registry validation (≤100 chars)
• Fixed PyPI workflow to handle re-tags gracefully (skip-existing)
• Chained MCP Registry publish after PyPI (prevents race condition)

From v0.3.3–v0.3.4 (merged)

• Dockerfile for containerized deployment
• PyPI trusted publishing workflow
• Pydantic models extracted to src/models.py
• SECURITY.md with responsible disclosure policy
• Smithery.ai configuration
• VS Code MCP launch config (.vscode/mcp.json)
• Various source code and test fixes

Full Changelog: v0.3.4...v0.4.1

v0.3.4 – Predictive Maintenance MCP Server: An open-source framework for integrating LLMs with predictive maintenance and fault diagnosis workflows

Abstract

Predictive Maintenance MCP Server is an open-source Model Context Protocol (MCP) server that bridges Large Language Models with industrial predictive-maintenance workflows. It exposes 21 specialist tools — vibration analysis (FFT, envelope spectrum), bearing-fault diagnostics (BPFO/BPFI/BSF/FTF), ISO 20816-3 severity classification, feature extraction, One-Class SVM anomaly detection, and automated HTML reporting — as first-class MCP resources consumable by any MCP-compatible AI client (Claude Desktop, VS Code Copilot, Cursor, etc.).

The server is self-contained: no external database, no cloud dependency, no GPU. It runs on a single machine with Python ≥ 3.11 and includes 20 production-quality bearing vibration signals (healthy baselines, inner race faults, outer race faults) sourced from the MathWorks RollingElementBearingFaultDiagnosis dataset.

Keywords: predictive maintenance, MCP, Model Context Protocol, vibration analysis, bearing fault diagnosis, LLM tool-use, Industry 4.0, ISO 20816, envelope analysis, condition monitoring, anomaly detection

---

What's new in 0.3.4

Added

• MCP Registry metadata — server.json + <!-- mcp-name --> tag in README for official registry listing at registry.modelcontextprotocol.io
• skip-existing flag on TestPyPI publish step to avoid 400 errors on re-upload

Changed

• Version bump to republish with updated README containing MCP Registry verification tag

Infrastructure

• .gitignore updated to exclude mcp-publisher CLI artifacts and token files
• Package successfully registered on the official MCP Registry as io.github.LGDiMaggio/predictive-maintenance-mcp

---

Install

pip install predictive-maintenance-mcp==0.3.4

v0.3.3 - First PyPI Release

Abstract

Predictive Maintenance MCP Server is a Model Context Protocol (MCP) server that brings industrial machinery diagnostics directly to Large Language Models such as Claude, enabling AI-powered vibration analysis, bearing fault detection, and predictive maintenance workflows through natural conversation.

The server exposes a comprehensive set of diagnostic tools — vibration data loading, FFT spectrum analysis, envelope analysis for bearing faults, ISO 20816-3 compliance evaluation, semi-supervised ML anomaly detection, and interactive HTML report generation — allowing LLM agents to orchestrate complex industrial diagnostics pipelines without requiring domain-specific coding from the end user.

This release (v0.3.3) addresses a critical packaging issue that prevented Python source files from being included in the published wheel, and aligns all project metadata, CI configuration, and citation files for production readiness.

What's new in v0.3.3

Fixed

• Critical: Fixed empty wheel — Python source files were not included in PyPI packages
• Explicit predictive_maintenance_mcp package mapping replacing broken packages.find
• Smart DATA_DIR resolution (PDM_PROJECT_DIR env var → cwd → __file__)
• License format updated to SPDX literal ("MIT") removing setuptools deprecation warnings

Changed

• CI Black line-length aligned to 88 (matching pyproject.toml)
• Removed premature --cov-fail-under=80 from pytest configuration
• Pydantic models extracted to dedicated models.py
• Added Dockerfile, SECURITY.md, GitHub issue/PR templates, Smithery manifest

Installation

pip install predictive-maintenance-mcp

Updated Metadata and Zenodo Integration

🔄 Metadata Update Release

This release updates project metadata and integrates proper Zenodo configuration for accurate DOI assignment.

🎯 What's New

Metadata Improvements

• ✅ Updated project title to better reflect the framework's purpose:
  • From: "A Proof of Concept"
  • To: "An open-source framework for integrating Large Language Models with predictive maintenance and fault diagnosis workflows"
• ✅ Added .zenodo.json - Ensures accurate metadata for Zenodo DOI assignment
• ✅ Enhanced documentation - Improved "Why This Matters" section to emphasize augmenting human engineering work

Documentation Updates

• 📚 Updated CITATION.cff with new title and version 0.3.2
• 📚 Updated README.md with Zenodo DOI badge and citation
• 📚 Improved messaging around LLM-augmented diagnostic workflows

Initial Public Release

🎉 Initial Public Release

✨ Features

• Real Data: 15 production bearing vibration signals included
• Advanced Diagnostics: FFT spectrum, envelope analysis, ISO 20816-3
• ML Anomaly Detection: Semi-supervised OneClassSVM/LOF with validation
• Professional Reports: Interactive HTML with Plotly visualizations
• Document Reader: Extract specs from equipment manuals
• Zero Configuration: Auto-detect sampling rates from metadata

🔧 MCP Server

• 25+ diagnostic tools for industrial machinery
• Hybrid architecture: Resources (data access) + Tools (processing)
• FastMCP-powered for Claude Desktop integration

📚 Documentation

• Complete API reference
• Step-by-step examples with real datasets
• Installation guide for Windows/macOS/Linux

🔬 Research

Note: Codebase generated using Claude AI under human supervision to explore MCP-based approaches for predictive maintenance workflows. Thorough testing required for production use.