The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/agent.py contains a function run_volatility_rootkit_scan which utilizes subprocess.run() with shell=True. The command string is constructed using f-string interpolation: f"vol3 -f {memory_dump} {plugin}". This implementation is vulnerable to command injection if the memory_dump or plugin arguments contain shell metacharacters.
[COMMAND_EXECUTION]: The instructions in SKILL.md guide the user to execute high-privilege commands including raw disk access (dd if=/dev/sda), file system mounting, and hardware-level firmware manipulation (flashrom, chipsec). While consistent with the malware analysis use case, these operations provide a significant attack surface for accidental or malicious system damage.
[PROMPT_INJECTION]: The skill exhibits an Indirect Prompt Injection surface (Category 8). It ingests untrusted binary data from boot sectors and firmware modules which is then processed by analysis tools (e.g., ndisasm, vol3). If the agent interprets the output of these tools, adversarial instructions embedded in the malware samples could influence the agent's logic.
Ingestion points: Raw data from dd, UEFIExtract, and vol3 analysis output entering the agent's context.
Boundary markers: None identified; analysis results are processed as plain text.
Capability inventory: Subprocess execution in agent.py, file system writes, and privileged CLI tool execution in SKILL.md workflow.
Sanitization: No evidence of input validation or escaping for the arguments passed to shell-executing functions.

analyzing-bootkit-and-rootkit-samples