The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from the SigNoz environment (view names, categories, and filters), which presents an indirect prompt injection surface. * Ingestion points: External view data is retrieved using signoz:signoz_list_views and signoz:signoz_get_view in SKILL.md. * Boundary markers: The instructions do not specify the use of delimiters to separate retrieved view content from agent instructions. * Capability inventory: The skill defines tools for creating, updating, and deleting views (signoz:signoz_create_view, signoz:signoz_update_view, signoz:signoz_delete_view) in SKILL.md. * Sanitization: No automated sanitization is applied to the retrieved view data before it is incorporated into the agent's context.
[SAFE]: The skill includes extensive human-in-the-loop safeguards that mitigate the risk of automated or malicious actions. * Mandatory Confirmation: All destructive operations and significant modifications require explicit user preview and confirmation as specified in SKILL.md. * Schema Validation: The skill mandates the use of a specialized sub-skill for query generation and validation, ensuring payloads conform to the expected v5 spec. * Trusted Resources: Documentation links and MCP tool calls target the official vendor domain (signoz.io) and infrastructure.

signoz-managing-views