The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documentation includes examples of using curl to query local Prometheus TSDB endpoints for diagnosing metric cardinality issues. These commands are intended for internal troubleshooting within a controlled environment.
[REMOTE_CODE_EXECUTION]: The instructions suggest using docker run to execute otelcol validate against local configuration files. This utilizes the official OpenTelemetry Collector images to verify syntax and component validity as part of a safe systems review workflow.
[CREDENTIALS_UNSAFE]: Example configurations in the evaluation files utilize environment variable interpolation (${env:APM_TOKEN}) for sensitive tokens, aligning with industry standards for secure secret management rather than hardcoding credentials.
[DATA_EXFILTRATION]: The skill explicitly defines rules and workflows for redacting PII, authorization headers, and other sensitive data within the telemetry pipeline using OpenTelemetry's native processors like transform and redaction.
[INDIRECT_PROMPT_INJECTION]: The skill prepares the agent to process external configuration and log data by providing strict rules for evidence-based conclusions, boundary markers for signal interpretation, and a requirement to name the specific questions telemetry must answer.

observability