The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides detailed instructions and commands for achieving Remote Code Execution (RCE) on target systems through various database features.
Evidence in references/mssql-attack.md: Instructions for using xp_cmdshell, OLE Automation, and SQL Server Agent Jobs to execute system commands.
Evidence in references/postgresql-attack.md: Instructions for using COPY FROM PROGRAM and creating UDFs/PL extensions for command execution.
Evidence in references/redis-attack.md: Instructions for Master-Slave replication RCE and Lua sandbox escape.
[PRIVILEGE_ESCALATION]: The skill includes explicit steps for escalating user privileges both within the database environment and to the underlying operating system.
Evidence in references/mssql-attack.md: Methods to escalate from db_owner to sysadmin and impersonate the sa account.
Evidence in references/postgresql-attack.md: Procedures for escalating to superuser and leveraging database access to gain root access on the host OS via sudo or crontab.
[PERSISTENCE]: The skill instructs the agent on establishing persistent access to target systems.
Evidence in references/redis-attack.md: Detailed methods for writing SSH public keys to the authorized_keys file and creating malicious entries in the system crontab via Redis.
[DATA_EXFILTRATION]: Techniques for capturing sensitive authentication data are described for theft.
Evidence in references/mssql-attack.md and references/postgresql-attack.md: Instructions for triggering UNC path access (SMB) via xp_dirtree, xp_fileexist, or lo_import to capture NTLM hashes using external tools like Responder.
[CREDENTIALS_UNSAFE]: The documentation provides lists of default and common weak credentials for numerous database services, facilitating automated brute-force attacks.
Evidence in SKILL.md, references/mssql-attack.md, and references/postgresql-attack.md: Lists of common passwords such as sa, 123456, admin@123, and Password1.

database-tactics