The Agent Skills Directory

[SAFE]: The skill is a legitimate collection of monitoring and observability configurations, rules, and scripts. It defines how to wire traces, collect metrics, and design dashboards using industry-standard tools.
[EXTERNAL_DOWNLOADS]: The skill references and encourages the use of well-known monitoring and observability services, including Prometheus, Grafana, OpenTelemetry, and Langfuse. It provides links to official documentation for these services.
[COMMAND_EXECUTION]: The skill includes shell command examples for Prometheus and Docker Compose setup, all of which are aligned with the intended purpose of setting up observability infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill defines patterns for processing and analyzing user content with LLMs (e.g., quality checks, summarization). While this introduces an attack surface for indirect prompt injection, the skill includes detection patterns such as 'statistical baseline anomaly detection' and 'LLM-as-judge' heuristics to identify and alert on malformed or malicious outputs.

monitoring-observability