forensics-checklist

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The skill’s required workflow uses the user-provided target argument ($ARGUMENTS → Step 1/3/5/6) to guide collection and “process forensic artifacts, log files, memory dumps, and system configuration data” that can include attacker-planted content at runtime, so outsider-authored text can be ingested indirectly via evidence/log contents without selecting a specific known item first.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The skill contains explicit privileged system-altering commands (e.g., "sudo insmod" to load a kernel module and disk imaging/snapshot commands) that instruct actions requiring root/IAM and would modify the host or target state.