AutoFTE

AutoFTE is a small local workflow I built while messing around with AFL++, crash triage, and basic binary analysis. It is not meant to be a polished platform. The goal is just to take a toy target from "I have crash files" to "I have a readable summary of what probably broke."

What it does

• Runs quick crash triage and groups similar inputs together
• Checks the target for common protections like NX, PIE, RELRO, and canaries
• Optionally asks a local Ollama model for a short write-up and patch ideas
• Generates a simple markdown summary and a static HTML dashboard

Repo layout

• vuln.c: intentionally vulnerable demo target
• fuzzer.sh: thin AFL++ wrapper for local runs
• triage.py: groups crash files by debugger frame or crash signal
• binary_analyzer.py: collects quick protection info
• llm_analyzer.py: optional Ollama-based notes
• report_builder.py: builds a short markdown summary
• sec_dash.py: writes a lightweight static dashboard
• pipeline.sh: runs the normal local workflow

Quick start

1. Create the micromamba env:

micromamba create -f environment.yml
micromamba activate autofte

2. Build the sample target:

make

3. Make sure you have at least one crash file in out/default/crashes/.

4. Run the pipeline:

./pipeline.sh ./target vuln.c

Outputs land in the repo root:

• crash_triage.json
• binary_analysis.json
• llm_analysis.json if Ollama is reachable
• analysis_summary.md
• dashboard/index.html

Notes

• AFL++, gdb, and the usual binutils tools still need to exist on the machine.
• If gdb is missing, crash grouping falls back to signal-based buckets.
• Ollama is optional. If it is not running, the rest of the pipeline still finishes.
• mythic_integrator.py is just an old side experiment and is not part of the normal flow anymore.