The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/agent.py script utilizes the subprocess.run function to execute the pacu command-line utility for AWS testing. The commands are constructed using list-based arguments, which follows security best practices to prevent shell injection.
[EXTERNAL_DOWNLOADS]: The SKILL.md file provides instructions to download and install the Pacu framework from its official GitHub repository (github.com/RhinoSecurityLabs/pacu). This is a well-known industry tool for cloud penetration testing.
[DATA_EXFILTRATION]: The scripts/agent.py script contains an export_session function that reads from the local Pacu SQLite database (which stores enumerated AWS resource data) and writes the content to a user-defined JSON file. While this involves data movement, it is a standard reporting feature for penetration testing tools.
[PROMPT_INJECTION]: The skill instructs the agent to process data enumerated from a live AWS account (e.g., S3 bucket names, IAM policy names). This represents an indirect prompt injection surface if the target environment contains maliciously named resources, though the script uses safe subprocess execution to mitigate local impact.

exploiting-aws-with-pacu