CTF Challenge for CSAW Finals 2021

Overview

Terminal Velocity

Misc CTF Challenge for CSAW Finals 2021

This is a challenge I've had in mind for almost 15 years and never got around to building until now. It (ab)uses a number of terminal escape codes to trigger both legitimate and scary and potentially dangerous terminal features, many of which are enabled by default in modern terminals! While a number of more serious exploits were patched in terminals since this original idea (it used to be trivial to kill many terminals with such escapes as "move cursor left 2^32 times or other similar ridiculous instructions), but most of the remaining shenanigans are merely abusing "legitimate" features that maybe are undesirable when simply viewing a text file or connecting to a network socket.

It's worth noting that because this service uses the usual netcat connection from clients, it will be line-buffered. This prevents some more egregious abuse of terminal escapes and requires some slight trickery to receive the return escape codes as hidden parts of existing responses during the various "press enter to continue" or similar prompts. Using something like telnet or ssh would fix this and allow for even more dangerous terminal manipulations.

Part of the goal of this challenge is to encourage people to be a bit more careful even when taking actions they might otherwise consider benign. Text should be considered harmful.

Deploying / Running

$ docker build -t terminal .
$ docker run -it -p 3535:3535 terminal

Or just run python3 service.py and connect to your local machine on port 3535.

Solutions

Two play-testers provided (partial) solutions which needs some tweaks for the final updates. This writeup covers the main tasks:

Level 0

When you first connect to the server you simply see a password (Level 0 Is Really Easy) which when pasted is indeed, correct.

The only tricky thing is if you try to sniff the connection or use a non-terminal to access it, you'll see that the password is originally something else that is overwritten.

Screen Check

After solving level 0, you will be asked "What is the proper screen size"? Some people may simply know the default terminal size is 80x24 and adjust accordingly, but if not, the server helpfully tells you whether your terminal was too big or too small after verifying that it can read your screen dimensions. If you're not running a real terminal, you'll need to learn to fake the correct response.

Level 1

Level one simply prints the password out in a black text on a black background. You can simply copy/paste it from the terminal but if you try to view it from the raw network traffic you will see that it is interspersed with unrelated escape codes that you have to filter out.

Correct pass: G1V3M3TH3N3XTL3V3L

Feature Check: Iconify

The next check will attempt to icnoify your terminal and query the status of the terminal (it also has the side effect of querying a user's iTerm current profile name). Correct approaches to solving this usually involve analyzing the query string that is sent and finding what is being looked up. Note that some VT100 references will give misleading answers and the oracle should always be consulted (if people get stuck I'd give this as a hint out since it's a better reference. All references are painfully hard to search though which is hilarious.)

Note: if you don't have a terminal capable of following this live (Terminal.app is the only one I know that does it all correctly) and you don't want to write a terminal emulator/manual interaction script (definitely the right approach for what's coming next), then you can work around it by just pressing backspace twice and entering 2t before pressing enter.

Level 2

Level two is similar except the line of correct text is erased after rendered. An emulator that simulates specific character drawing will be able to recover the text, or a filter that blocks the "erase" escape codes (though there are several used) can work here.

Correct password: HalfwayDone

Level 3

Level three is pretty nasty (the user is warned though!)

It will attempt to do all sorts of nasty things to their terminal including printing locking the prompt, crashing it with bogus operations. One printer accidentally spewed out paper and one Windows machine blue screened during the testing of these features, so this can be tough! (The printer bug was fixed with all the aforementioned iTerm detection above ). By this point users should be strongly considering not directly interacting with the port but using pwntools with heavy filtering or some other method. (Fun fact, Windows terminal actually looks super robust against these sorts of shenanigans and the developers even built an entire fuzzing harness that really needs to be run against all other major browsers which still have many bugs).

Anyway, for players who have been building a very simply terminal emulator to this point, just having the ability to emulate three different cases of move and draw commands will let them re-create the correct text for this level.

Correct password: BobTheBuilder

Level 4

The final level brings image formats! Yup, there are actually many different valid forms of images that can be displayed in terminals. Though, if I've done my job correctly, the previous level will have broken or rendered useless most of the terminals that otherwise could just show the images directly.

The three images that are displayed are:

  1. Simple base64 encoded file in iTerm image format
  2. A sixel and
  3. A Tektronix image

There are several different approaches/tools to solving the last two images. Just using a compatible terminal and separately cat'ing the file after extracting them from the session is sufficient.

xTerm has the only support for the final Tektronix image format I have found. When assembled, the images reveal the final passcode: PINEY_FLATS_TN_USA a random city with no meaning at all behind it.

Entering the final password reveals the flag for the challenge!

Hopefully people have a lot more respect about what their terminals are capable of after working on this challenge and maybe even take more care when randomly connecting to servers on the internet.

Owner
Jordan
Jordan
Out-of-Domain Human Mesh Reconstruction via Dynamic Bilevel Online Adaptation

DynaBOA Code repositoty for the paper: Out-of-Domain Human Mesh Reconstruction via Dynamic Bilevel Online Adaptation Shanyan Guan, Jingwei Xu, Michell

198 Dec 29, 2022
Source code for the NeurIPS 2021 paper "On the Second-order Convergence Properties of Random Search Methods"

Second-order Convergence Properties of Random Search Methods This repository the paper "On the Second-order Convergence Properties of Random Search Me

Adamos Solomou 0 Nov 13, 2021
Code for NeurIPS2021 submission "A Surrogate Objective Framework for Prediction+Programming with Soft Constraints"

This repository is the code for NeurIPS 2021 submission "A Surrogate Objective Framework for Prediction+Programming with Soft Constraints". Edit 2021/

10 Dec 20, 2022
Code for the submitted paper Surrogate-based cross-correlation for particle image velocimetry

Surrogate-based cross-correlation (SBCC) This repository contains code for the submitted paper Surrogate-based cross-correlation for particle image ve

5 Jun 30, 2022
Multilingual Image Captioning

Multilingual Image Captioning Authors: Bhavitvya Malik, Gunjan Chhablani Demo Link: https://huggingface.co/spaces/flax-community/multilingual-image-ca

Gunjan Chhablani 32 Nov 25, 2022
🗣️ Microsoft Edge TTS for Home Assistant, no need for app_key

Microsoft Edge TTS for Home Assistant This component is based on the TTS service of Microsoft Edge browser, no need to apply for app_key. Install Down

152 Dec 31, 2022
deep learning model that learns to code with drawing in the Processing language

sketchnet sketchnet - processing code generator can we teach a computer to draw pictures with code. We use Processing and java/jruby code paired with

41 Dec 12, 2022
A machine learning library for spiking neural networks. Supports training with both torch and jax pipelines, and deployment to neuromorphic hardware.

Rockpool Rockpool is a Python package for developing signal processing applications with spiking neural networks. Rockpool allows you to build network

SynSense 21 Dec 14, 2022
PyTorch implementation of ARM-Net: Adaptive Relation Modeling Network for Structured Data.

A ready-to-use framework of latest models for structured (tabular) data learning with PyTorch. Applications include recommendation, CRT prediction, healthcare analytics, and etc.

48 Nov 30, 2022
Code for generating the figures in the paper "Capacity of Group-invariant Linear Readouts from Equivariant Representations: How Many Objects can be Linearly Classified Under All Possible Views?"

Code for running simulations for the paper "Capacity of Group-invariant Linear Readouts from Equivariant Representations: How Many Objects can be Lin

Matthew Farrell 1 Nov 22, 2022
DeepOBS: A Deep Learning Optimizer Benchmark Suite

DeepOBS - A Deep Learning Optimizer Benchmark Suite DeepOBS is a benchmarking suite that drastically simplifies, automates and improves the evaluation

Aaron Bahde 7 May 12, 2020
A repo that contains all the mesh keys needed for mesh backend, along with a code example of how to use them in python

Mesh-Keys A repo that contains all the mesh keys needed for mesh backend, along with a code example of how to use them in python Have been seeing alot

Joseph 53 Dec 13, 2022
NDE: Climate Modeling with Neural Diffusion Equation, ICDM'21

Climate Modeling with Neural Diffusion Equation Introduction This is the repository of our accepted ICDM 2021 paper "Climate Modeling with Neural Diff

Jeehyun Hwang 5 Dec 18, 2022
Implementation of 'lightweight' GAN, proposed in ICLR 2021, in Pytorch. High resolution image generations that can be trained within a day or two

512x512 flowers after 12 hours of training, 1 gpu 256x256 flowers after 12 hours of training, 1 gpu Pizza 'Lightweight' GAN Implementation of 'lightwe

Phil Wang 1.5k Jan 02, 2023
Code repository for the paper "Doubly-Trained Adversarial Data Augmentation for Neural Machine Translation" with instructions to reproduce the results.

Doubly Trained Neural Machine Translation System for Adversarial Attack and Data Augmentation Languages Experimented: Data Overview: Source Target Tra

Steven Tan 1 Aug 18, 2022
A python code to convert Keras pre-trained weights to Pytorch version

Weights_Keras_2_Pytorch 最近想在Pytorch项目里使用一下谷歌的NIMA,但是发现没有预训练好的pytorch权重,于是整理了一下将Keras预训练权重转为Pytorch的代码,目前是支持Keras的Conv2D, Dense, DepthwiseConv2D, Batch

Liu Hengyu 2 Dec 16, 2021
Earth Vision Foundation

EVer - A Library for Earth Vision Researcher EVer is a Pytorch-based Python library to simplify the training and inference of the deep learning model.

Zhuo Zheng 34 Nov 26, 2022
scikit-learn: machine learning in Python

scikit-learn is a Python module for machine learning built on top of SciPy and is distributed under the 3-Clause BSD license. The project was started

scikit-learn 52.5k Jan 08, 2023
A general, feasible, and extensible framework for classification tasks.

Pytorch Classification A general, feasible and extensible framework for 2D image classification. Features Easy to configure (model, hyperparameters) T

Eugene 26 Nov 22, 2022
Source code for Task-Aware Variational Adversarial Active Learning

Contrastive Coding for Active Learning under Class Distribution Mismatch Official PyTorch implementation of ["Contrastive Coding for Active Learning u

27 Nov 23, 2022