Home

Proc_Map_Analyzer

Interactive 3D process graph analyzer with deterministic risk scoring and local AI enrichment (Ollama) — no data leaves your machine.

Proc_Map_Analyzer analyzes the processes running on a machine, their related files (executable, working directory, open files) and their relationships (parent → child, shared files, network connections), computes a risk level through deterministic rules, optionally enriches each significant process via a local Ollama model, and produces an interactive 3D graph you can explore in any browser.

Quick start

git clone https://github.com/TFD-42/Proc_Map_Analyzer.git
cd Proc_Map_Analyzer
./install.sh        # macOS / Linux / Termux — installs everything, then runs

Or, if Python 3.9+ is already installed:

python3 process_analyzer_allinone.py

Launched without arguments, the script starts an interactive wizard (2–4 questions) and opens the resulting 3D graph in your browser. Launched with options, it behaves like a classic CLI tool.

Wiki pages

Page	What you'll find
Installation	Automatic installers (macOS/Linux/Windows/Termux), prerequisites, manual setup
Getting Started	Wizard mode, first run, output files
CLI Reference	Every command-line flag, with defaults
The 3D Graph	Display modes, legend, keyboard shortcuts, copy buttons
Risk Engine	The deterministic rules, AI escalation, whitelist/blacklist config
Advanced Modes	Watch, forensics (--pid), history/compare, baseline, integrity, cache, sandbox, plugins
Ollama Integration	Model selection per device, preload, reliability tuning
Android / Termux	The /proc backend, mini model, platform limitations
Building an Executable	PyInstaller build, mandatory flags, Gatekeeper note
Privacy and Security	What talks to the network (almost nothing), reporting vulnerabilities
Troubleshooting	Symptom → cause → fix table

Key facts

• Cross-platform: macOS, Linux, Windows (PyInstaller-compilable) and Android/Termux (built-in /proc backend).
• Local-first: the analysis never leaves your machine; the only external call is the 3D rendering library CDN when the HTML opens.
• AI is optional: without Ollama, the deterministic rule engine still produces named, explainable risk signals.
• License: MIT. Current release: v0.1.0.