[DEPREICATED] Moved to new org https://github.com/haKC-ai/hakcthropic/

This repo just aims to get you started with Anthropics Quickstarts environment to deploy "AI" hacking agents for shenanigans.

Why: On Oct 22, 2024 Anthropic released and in part it states:

"We’re also introducing a groundbreaking new capability in public beta: computer use. Available today on the API, developers can direct Claude to use computers the way people do—by looking at a screen, moving a cursor, clicking buttons, and typing text. Claude 3.5 Sonnet is the first frontier AI model to offer computer use in public beta."

So in this repo, I am showing how the install guide leverages this to install metasploit, set options and execute an attack.

[DEPREICATED] Moved to new org https://github.com/haKC-ai/hakcthropic/

• on the VM in as the user home dir, there is a hidden directory called ~/.anthropic/ which I found two files:

  • api_key

  • system_prompt

    • The default state of the system prompt is blank, I had decent luck giving it instructions similar to "jailbreaks".

    • Documentation for this is here: https://docs.anthropic.com/en/docs/build-with-claude/computer-use

      • I had pretty good luck with it respecting these prompts prior to running the commands issues in the streamlit input field
      • 

      Interesting note: Even though its getting instruction to not intereact with external resources, it clearly ignores them. ya know.. since I was able to clone MSF and run it against something externally.

      

• I experienced this issue "Claude sometimes assumes outcomes of its actions without explicitly checking their results. " with some of my commands dispite telling it:

  ... Run each command one at a time and make sure they complete. I want to see the output as you run the command. [DEPREICATED] Moved to new org https://github.com/haKC-ai/hakcthropic/

• Read the docs
• Get your Anthropic API key from the console

1. Clone this repository:

   git clone https://github.com/anthropics/anthropic-quickstarts.git
   #Then download my start_hacking.sh script here: https://github.com/NoDataFound/hackGPT/tree/main/hackthropic or just clone this entire repo
   git clone https://github.com/NoDataFound/hackGPT.git

2. Run the start_hacking.sh script:

   hackGPT/hackthropic/start_hacking.sh #or whereever you saved it

Add your ANTHROPIC_API_KEY API key to .env

Format of the .env file:

ANTHROPIC_API_KEY=<your_api_key>

The start_hacking.sh script will:

1. Create a Python virtual environment.
2. Install the required dependencies.
3. Export environment variables from the .env file.
4. Run the Docker container with appropriate port bindings and environment variables.

• Ensure Docker is installed and running on your system.
• The script drops the .env file in anthropic-quickstarts/computer-use-demo/.