The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests untrusted incident descriptions and event logs to generate structured reports. While this constitutes an indirect prompt injection surface, the risk is negligible as the provided scripts only perform text processing and have no capabilities for network exfiltration, arbitrary command execution, or destructive file system operations. Evidence: Ingestion occurs in incident_classifier.py and timeline_reconstructor.py via JSON input. Boundary markers are absent in raw templates but reports are highly structured. Capabilities are limited to stdout and argument-defined file output. Sanitization relies on standard JSON parsing.

incident-commander