The Agent Skills Directory

[COMMAND_EXECUTION]: The skill interacts with Prometheus servers using standard system utilities (curl, jq, date, sed). All inputs, including PromQL queries and time parameters, are sanitized using jq --arg or curl --data-urlencode and validated with regular expressions to prevent command injection.- [CREDENTIALS_UNSAFE]: Sensitive bearer tokens are managed through the PROMETHEUS_TOKEN environment variable. The skill ensures these credentials are only transmitted via HTTP headers and never logged, printed to standard output, or included in JSON responses, following secure secret management protocols.- [SAFE]: The skill is strictly read-only and connecting only to user-defined Prometheus instances. It performs scheme validation (http/https only) and implements safe temporary file management with proper cleanup traps, minimizing risks of unauthorized access or file system exposure.

prom-query