signoz-managing-views
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied intent to generate API payloads for creating, updating, and deleting views in SigNoz. This presents a potential surface for indirect injection if a user attempts to save malicious instructions.
- Ingestion points: View names, display labels, and filter intent are accepted from the user as described in
SKILL.md. - Boundary markers: The skill requires an explicit preview and user confirmation (for human-in-the-loop) or detailed logging (for autonomous agents) before performing any write operations. It explicitly warns against fabricating payloads.
- Capability inventory: The skill uses
signoz_create_view,signoz_update_view, andsignoz_delete_viewtools, which perform authenticated write actions on the SigNoz platform. - Sanitization: The skill effectively mitigates injection risks by delegating query construction to a specialized validation skill (
signoz-generating-queries) and performing a mandatory 'sample fetch' to verify the exact filter against live data before persistence, ensuring the view matches expected data signal rules.
Audit Metadata