A GUI front-end for the RuleFlow pipeline: rulest + concentrator + ranker, all wired together to get the most out of the extracted Hashcat rules.

What it actually does

You feed it three files—a base wordlist, a target wordlist, and cracked passwords. You select an OpenCL device, choose a preset, and you can adjust the expected runtime (higher values ​​result in more chains generated), click Run, and the program runs through the entire chain: informed extraction, genetic rule extraction, rule cleaning/optimization, and then scoring and ranking the survivors from the larger results. Finally, you get a structured CSV file.

The presets

Mode	       Generations	Populations	Best for
Fast	       150	        300	        Quick tests, medium dictionaries
Balanced       300	        600	        Daily driver, good default
Maximum	       500	        1000	        Deep runs, large or stubborn sets

Why I built this

The backend tools are solid, but running them one-by-one with long command lines gets old fast. This wraps the whole flow in a single window with:

• Live log with ANSI colors, horizontal scrolling for long lines, and auto-trim so it doesn't choke on verbosity
• Memory watchdog — monitors RAM/VRAM usage and auto-suggests switching to Low-Memory preset if a stage gets killed by OOM
• OpenCL device scanner — detects your GPU or falls back to CPU without manual --device flags
• MAB vs Legacy ranker — pick adaptive bandit sampling for huge skewed rule sets, or exhaustive scoring when you want full statistical accuracy

Quick start

git clone https://github.com/A113L/ruleflow.git
cd ruleflow
python rcr

Requires Python 3 + tkinter. The backend scripts (rulest_v2.py, concentrator.py, ranker.py) should be in the same directory.

Feedback or bug reports welcome — this is a personal tools that grew a UI, so rough edges are expected.