forensics-checklist

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime workflow can ingest attacker-controlled free text from outsider-authored log files / forensic artifacts (e.g., Step 5 “Log Preservation” and Step 3 “Volatile Data Capture” imply exporting/processing logs and evidence that may contain adversarial strings), which the agent then includes in the LLM context for report generation—an indirect prompt-injection risk.

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

• Attempt to modify system services in skill instructions detected (high risk: 1.00). This skill contains explicit privileged commands that modify system state (e.g., "sudo insmod" to load LiME, raw disk imaging of /dev/sda, use of write-blockers and snapshots) which require elevated privileges and could change kernel or disk state, so it pushes the agent toward actions that would compromise the host environment.