The Agent Skills Directory

[SAFE]: No security issues detected. The skill is designed for observability and adheres to secure patterns for tool development, such as requiring explicit discovery before querying and using placeholders for authentication credentials.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection due to its core function of retrieving and processing untrusted observability data (logs and traces) from external Kubernetes clusters. However, this risk is inherent to the primary purpose of the skill and is mitigated by tool design. 1. Ingestion points: Untrusted telemetry data enters the agent context via the search-logs and search-traces tools. 2. Boundary markers: No specific delimiters are enforced within the prompt text for tool outputs. 3. Capability inventory: The skill includes specific write operations for creating resources like alerts and dashboards. 4. Sanitization: All write-annotated tools (create-alert, create-dashboard) require human-in-the-loop approval at the agent platform level before execution.

kubesense-skills