The Agent Skills Directory

[COMMAND_EXECUTION]: The references/tools.md file defines an execute_command function designed to run tools not exposed as specific MCP endpoints. This provides the agent with a general-purpose shell execution capability, which can be exploited to run unauthorized commands on the host system if input is not strictly constrained.\n- [INDIRECT_PROMPT_INJECTION]: The skill's primary workflow involves processing untrusted external files, creating a surface for indirect prompt injection attacks.\n
Ingestion points: Forensic artifacts such as memory dumps (volatility_analyze), network captures, and image files (exiftool_extract, steghide_analysis) are passed to the agent for processing.\n
Capability inventory: The agent has access to arbitrary command execution via execute_command, file system write operations through foremost_carving, and various data extraction utilities.\n
Boundary markers: There are no instructions provided to the agent to treat data from forensic artifacts as untrusted or to ignore potential instructions embedded within them.\n
Sanitization: The skill lacks explicit sanitization or validation logic for arguments passed to forensic tools, particularly in the additional_args parameters and the raw command string in execute_command.

pentest-ctf-forensics