function tvx() {<repo-root>/tvx.py $@}

pip install -r ./requirements.txt
sudo apt-get update
sudo apt-get install build-essential cmake tmux cpuid linux-tools-common

tvx <command> <subcommand>

# start the fuzzer
tvx run start

# clear the progress made and all results
tvx run cleanup

# classify the results 
# (typically the first thing you want to do after fuzzing)
tvx result classify

# view the content of a testfile
tvx result view <result-file.json>

# export a given test file into a standalone 'reproducer'
# allows you to inspect the result further
tvx result export <result-file.json>

# rerun the testcase inside the fuzzer
tvx result rerun

# install/check Trevex system dependencies (apt and python packages)
tvx setup install

# load the tvx Python environment
tvx setup load

uarch-lab01  # my server
uarch-lab03  # my dev machine 
uarch-lab07  # my other dev machine

tvx ctrl -m <machine_file.cfg> <subcommand>

# prepare the machines for the fuzzing campaign
# ATTENTION: this might change the running kernel and reboot the machine
#   This is not always needed, you can just try skipping the step.
tvx ctrl -m ./my-servers.cfg setup

# Start Trevex on the remote machines and attach to their tmux sessions
# Note: If Trevex fails to start, the skipped setup step might be the reason.
tvx ctrl -m ./my-servers.cfg spawn

# Detach from all tmux sessions and let it run for a while
tvx ctrl -m ./my-servers.cfg detach

# Attach again to the tmux sessions
tvx ctrl -m ./my-servers.cfg attach

# Stop all Trevex instances once you're done
tvx ctrl -m ./my-servers.cfg stop

# Retrieve the results and store them on your local machine
# Note: While this is often useful for organization, actual 
#   reproduction steps should executed on the *exact same* 
#   CPU that was fuzzed.
tvx ctrl -m ./my-servers.cfg pull-results

# Reset the state on all machines. This cleans all progress
# made and deletes all results.
tvx ctrl -m ./my-servers.cfg cleanup

@inproceedings{Weber2026Trevex,
 author = {Weber, Daniel and Thomas, Fabian and Trampert, Leon and Zhang, Ruiyi and Schwarz, Michael},
 booktitle = {{IEEE S\&P}},
 title = {{Trevex: A Black-Box Detection Framework For Data-Flow Transient Execution Vulnerabilities}},
 year = {2026}
}