The Agent Skills Directory

Credential Management: The skill uses a secure approach to authentication by relying on runtime-injected environment variables (GRAFANA_TOKEN) and explicitly advises against manual token management within the agent's logic. This minimizes the risk of credential exposure.
Network Operations: Interactions are restricted to the user-provided Grafana instance URL using standard tools like curl. There are no indications of data exfiltration to unauthorized third-party domains.
System Interactions: The skill utilizes common utilities such as jq for data parsing and date for timestamp generation. No high-risk commands or privilege escalation patterns were detected.
External Data Ingestion: While the skill processes data from external Grafana instances (such as dashboard content or alert rules), this is inherent to its purpose. Standard large language model guardrails typically manage the risks associated with processing such content.

grafana-api