The Agent Skills Directory

[SAFE]: The skill serves as a legitimate technical resource for DevOps tasks, providing standard boilerplate for JSON logging, Prometheus metrics collection, and service health monitoring.
[SAFE]: Includes explicit instructions to avoid logging sensitive information (PII, tokens, passwords), which is a critical security best practice for observability systems.
[SAFE]: The external utilities and libraries referenced, such as jq, journalctl, docker, and the prometheus_client Python package, are standard, well-known tools for system administration and monitoring.
[PROMPT_INJECTION]: The skill facilitates the analysis of external log data, which represents a potential surface for indirect prompt injection.
Ingestion points: Log analysis via jq and Python-based log parsing.
Boundary markers: None explicitly defined in the provided code snippets to delimit untrusted log content.
Capability inventory: Shell access for jq and journalctl, and Python execution for processing data.
Sanitization: No specific filtering or sanitization of log content is demonstrated in the boilerplate examples. This is considered a low-risk surface inherent to the skill's primary purpose.

monitor