The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill represents a potential attack surface for indirect prompt injection as it ingests and processes data from external Grafana instances via gcx command outputs.
Ingestion points: Data enters the context through gcx slo definitions get, gcx slo definitions timeline, gcx slo definitions status, and gcx metrics query outputs (SKILL.md).
Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands within the fetched metric metadata or SLO names.
Capability inventory: The skill has access to the Bash tool to execute gcx commands.
Sanitization: There is no explicit sanitization of external strings (like SLO names or label values) before they are processed for trend analysis and recommendation generation.
Risk Mitigation: The risk is minimal as the skill is explicitly advisory and instructs the agent to route the user to a separate slo-manage skill for any actual changes, providing a human-in-the-loop checkpoint.

slo-optimize