The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a shell script at scripts/collect-evidence.sh that executes system commands for evidence gathering, specifically kubectl get pods, journalctl, docker events, and top. These are used for read-only status collection.- [COMMAND_EXECUTION]: The scripts/collect-evidence.sh script implements a validation check on the INCIDENT_ID argument using the regular expression ^[A-Za-z0-9_-]+$, which mitigates the risk of command injection through the incident identifier.- [PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as its primary function involves ingesting and analyzing untrusted external data (logs, traces, and metrics). While no specific boundary markers or sanitization for log content are defined in the instructions to prevent the agent from obeying instructions embedded in logs, this is an inherent risk of the skill's analytical purpose.

production-incident-investigator