LGDiMaggio/mcp-motor-current-signature-analysis: v0.1.3

Abstract mcp-server-mcsa is an open-source server implementing the Model Context Protocol (MCP) for Motor Current Signature Analysis (MCSA) — a non-invasive condition-monitoring technique that detects mechanical and electrical faults in electric motors by analysing the harmonic content of stator-current signals. The server exposes 21 tools, 1 resource, and 1 prompt that enable any MCP-compatible AI assistant (e.g. Claude, GitHub Copilot) to perform the complete MCSA diagnostic workflow: signal acquisition from standard file formats (CSV, WAV, NumPy), preprocessing (DC removal, windowing, filtering), spectral analysis (FFT, Welch PSD, peak detection), envelope analysis (Hilbert transform demodulation), time-frequency analysis (STFT), and automated fault detection with severity classification for broken rotor bars, air-gap eccentricity, stator inter-turn short circuits, and bearing defects. Fault indices are computed from sideband amplitude ratios relative to the supply fundamental and classified into four severity levels (healthy, incipient, moderate, severe) based on thresholds from peer-reviewed literature and international standards. What's new in v0.2.0 Persistent data store — signals and spectra are persisted to disk (~/.mcsa_data/) as compressed NumPy .npz files, referenced by short IDs (sig_xxxx, spec_xxxx). This keeps large arrays out of the LLM chat context while preserving full data fidelity for all computations. Data survives server restarts. 2 new tools: list_stored_data, clear_stored_data All 19 existing tools updated to accept signal_id / spectrum_id parameters Fully backward compatible with raw array inputs Updated documentation (README, Usage Guide)