The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/agent.py script executes PowerShell commands via subprocess.run to query Windows Security event logs for detection purposes.
[COMMAND_EXECUTION]: The references/api-reference.md file provides a Python code snippet that uses subprocess.run with variable arguments (modules, output_dir), creating a potential command injection surface if the inputs are not properly validated before execution.
[DATA_EXFILTRATION]: The skill's primary purpose is to identify, extract, and prioritize sensitive credentials (passwords, session cookies, and system keys). scripts/process.py is designed to parse and organize these credentials from tool output files.
[EXTERNAL_DOWNLOADS]: The documentation provides instructions to download the LaZagne tool and its dependencies from a public GitHub repository (https://github.com/AlessandroZ/LaZagne.git).
[DATA_EXFILTRATION]: Indirect prompt injection surface identified in scripts/process.py which processes untrusted JSON data from tool outputs.
Ingestion points: load_lazagne_output in scripts/process.py reads data from a user-specified JSON file.
Boundary markers: Absent. The script processes raw dictionary keys and values from the JSON structure without delimiters.
Capability inventory: File read/write operations and terminal output printing across all scripts.
Sanitization: Absent. Values extracted from the JSON are directly printed to the console and written to report files.

performing-credential-access-with-lazagne