Invicti-Security/brainstorm — reverse-engineered prompt

Build me a Python command line security tool for authorized web testing that makes directory and file discovery smarter by combining ffuf with a local Ollama model.

I want to pass in a normal ffuf command with FUZZ in the URL, then the tool should run fuzzing cycles, collect discovered links, ask the local model to suggest better paths based on what was found, fuzz those suggestions, and keep repeating. Save all discovered paths to a file and show useful progress in the terminal.

Also include a second command for short filename discovery, where I can give it a target short name or filename and it tries to find related legacy style file paths. Let me choose the number of cycles, the Ollama model, success status codes, prompt file, and debug mode from command line options.

Add a simple benchmark script that can compare local models and produce a readable report. Assume ffuf and Ollama are installed locally, but make setup and usage clear in a README.