The Agent Skills Directory

[DATA_EXFILTRATION]: The skill provides methodologies for automated data extraction, including dumping entire databases and specifically targeting credential columns (e.g., username, password) using flags like --dump, --dump-all, and -C. It also facilitates the reading of sensitive system files, specifically citing /etc/passwd via the --file-read parameter.
[COMMAND_EXECUTION]: The skill instructs the agent on how to execute operating system-level commands on a remote target server using the --os-shell and --os-cmd parameters, which allows for full command execution capabilities once a vulnerability is exploited.
[PRIVILEGE_ESCALATION]: Includes guidance on leveraging database vulnerabilities to gain administrative access to the underlying operating system, specifically mentioning that techniques like --os-shell require DBA-level privileges.
[PERSISTENCE]: Provides instructions for the --file-write and --file-dest flags, which allow an attacker to upload arbitrary files (such as web shells or backdoors) to the target server's web root to maintain access.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data which could be manipulated by an attacker to influence agent behavior.
Ingestion points: Target URLs, HTTP request files captured from proxies (via -r), and bulk scan lists (via -m).
Boundary markers: None identified; the instructions do not include delimiters or warnings to ignore instructions embedded in the target data.
Capability inventory: Includes filesystem read/write, network operations to arbitrary targets, and OS command execution through the sqlmap tool.
Sanitization: The skill lacks any input validation or sanitization routines for the parameters passed to the command-line interface.

SQLMap Database Penetration Testing