The Agent Skills Directory

[Indirect Prompt Injection] (SAFE): The incident_triage.py script ingests incident descriptions via the --description command-line argument. While this data influences team assignment, the logic is limited to simple keyword matching, posing no risk of safety bypass or command injection.\n
Ingestion points: args.description in scripts/incident_triage.py.\n
Boundary markers: Not applicable (keyword matching logic used).\n
Capability inventory: Script allows writing a JSON report to a local file path via the --output argument.\n
Sanitization: Inputs are normalized to lowercase for matching purposes.\n- [Privilege Escalation] (SAFE): The documentation in references/troubleshooting.md mentions chmod for script execution and sudo for resolving permission issues. The inclusion of an explicit 'not recommended' warning for sudo aligns with secure administrative documentation practices.

incident-responder