The Agent Skills Directory

[COMMAND_EXECUTION]: Provides specific shell commands for performing SSRF attacks, password spraying, and interacting with administrative interfaces (OWA/ECP) to detect and exploit vulnerabilities.
[REMOTE_CODE_EXECUTION]: References the execution of unverified external Python exploit scripts for critical Exchange vulnerabilities (e.g., ProxyLogon, ProxyShell) to gain remote code execution.
[DATA_EXFILTRATION]: Outlines procedures for harvesting sensitive information, including Global Address Lists and full user mailboxes via PST export commands.
[PRIVILEGE_ESCALATION]: Includes methodologies for compromising domain-level accounts by exploiting Exchange's high-privilege permissions in the domain and using NTLM relaying to LDAP.
[PERSISTENCE]: Details the deployment of webshells to maintain access and provides methods to clear ECP activity logs to evade security monitoring.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input such as target URLs and user lists without validation, creating a surface for injection.
Ingestion points: Target hostnames and credential lists provided as variables in shell commands.
Boundary markers: None present.
Capability inventory: Subprocess execution (Python/PowerShell), network operations (curl/nmap), and file system access.
Sanitization: No evidence of input escaping or validation before interpolation into commands.

exchange-attack