Local, full-logit AI text forensics and humanization toolkit. It scores markdown with observer/performer llama.cpp models (logPPL, logXPPL, B), surfaces chunk-aware hotspots and heatmaps, ranks rewrites, and ships CLI, desktop UI, Obsidian plugin, VS Code extension, and HTTP API (dev-grade) server with GUI test/demo harness.
vscodevscode-extensionhumanizerobsidianstylometrycross-entropytkinter-guiperplexityhumanizationobsidian-pluginperplexity-scorebinocularsapi-httpgptzeroai-detectionllama-cpp-pythonlikelihood-scoringhumanizing-interactiveai-text-forensicsauthorship-signals
-
Updated
Feb 27, 2026 - Python