Skip to content

README.md

Latest commit

 

History

History
87 lines (60 loc) · 3.99 KB

README.md

File metadata and controls

87 lines (60 loc) · 3.99 KB

Predictive Maintenance Plugin for Claude Code

Industrial-grade predictive maintenance skills, agents, and workflows for Claude Code. Works with the predictive-maintenance-mcp server.

Prerequisites

The predictive-maintenance-mcp MCP server must be installed and connected. See the installation guide for setup instructions.

Installation

From Marketplace

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

This repo also includes .claude-plugin/marketplace.json at repository root for hosted distribution (owner/repo).

Local Development

/plugin marketplace add .
/plugin install predictive-maintenance@predictive-maintenance-marketplace

# Optional: test only the plugin subdirectory marketplace
/plugin marketplace add ./plugin
/plugin install predictive-maintenance@predictive-maintenance-marketplace

Components

Skills (7)

Skills activate automatically when Claude detects relevant context.

Skill Trigger Examples Description
bearing-diagnosis "diagnose bearing", "outer race fault" Full bearing fault diagnostic workflow
gear-diagnosis "gear fault", "gear mesh frequency" Gear fault detection via spectral analysis
quick-screening "quick check", "is this machine healthy" Fast health screening (<30s)
report-generation "generate report", "create PDF" Professional HTML/DOCX report generation
anomaly-detection "train model", "detect anomalies" ML-based anomaly detection (SVM, LOF)
signal-management "load signal", "generate test signal" Signal loading, generation, cache management
documentation-search "search manual", "look up bearing" RAG search across manuals and catalogs

Agents (2)

Agents run autonomously for complex, multi-step tasks.

Agent When to Use Description
diagnostic-pipeline "Run a full diagnostic on this signal" Complete ISO 13374 pipeline: load -> analyze -> diagnose -> report
signal-explorer "Explore these signals and compare them" Signal characterization, comparison, and outlier detection

Commands (3)

Quick entry points for common workflows.

Command Usage Description
/pm-diagnose /pm-diagnose bearing_001 Start bearing fault diagnosis
/pm-screen /pm-screen bearing_001 Quick health screening
/pm-report /pm-report bearing_001 full Generate diagnostic reports

MCP Tool Coverage

This plugin provides domain expertise for all 38 tools in the predictive-maintenance-mcp server:

  • Signal Acquisition: load_signal, generate_test_signal, list_signals, list_stored_signals, get_signal_info, clear_signal, clear_all_signals
  • Spectral Analysis: analyze_fft, analyze_envelope, compute_power_spectral_density, compute_spectrogram_stft, extract_features_from_signal
  • Visualization: plot_signal, plot_spectrum, plot_envelope, plot_iso_20816_chart
  • Bearing Diagnostics: calculate_bearing_characteristic_frequencies, check_bearing_fault_peak_tool, check_bearing_faults_direct, diagnose_vibration_tool, search_bearing_catalog, lookup_bearing_and_compute_tool
  • ISO Standards: evaluate_iso_20816, assess_vibration_severity
  • ML / Anomaly Detection: train_anomaly_model, predict_anomalies
  • Documentation: search_documentation, read_manual_excerpt, extract_manual_specs, list_machine_manuals, load_and_validate_metadata
  • Reports: generate_fft_report, generate_envelope_report, generate_iso_report, generate_feature_comparison_report, generate_pca_visualization_report, generate_diagnostic_report_docx

Standards Compliance

  • ISO 13374 — Six-block diagnostic architecture
  • ISO 20816-3 — Vibration severity classification
  • MIMOSA OSA-CBM — Condition-based maintenance framework

License

MIT