The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for executing high-privilege and security-relevant CLI tools including 'apt install', 'searchsploit', 'nmap', and 'chmod'.
[EXTERNAL_DOWNLOADS]: Documents methods for transferring files between systems using tools such as 'wget', 'certutil', and PowerShell's 'Invoke-WebRequest', which can be used to move malicious payloads.
[DYNAMIC_EXECUTION]: Details the compilation of C source code into executable binaries using 'gcc', including instructions for using flags that disable memory protections (e.g., '-fno-stack-protector' and '-z execstack') which are common in exploit development.
[INDIRECT_PROMPT_INJECTION]: The skill defines workflows for processing untrusted external data, such as Nmap XML scan results and third-party exploit scripts, creating a potential attack surface for indirect injection.
Ingestion points: Nmap scan output files ('scan.xml') and mirrored exploit code ('.c', '.py').
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat external file content as untrusted.
Capability inventory: Includes shell command execution, C compilation, package installation, and file permission modification.
Sanitization: The instructions rely on manual human inspection of exploit headers; no automated sanitization or verification processes are described.

searchsploit