dt-obs-network-flows

Pass

Audited by Gen Agent Trust Hub on Aug 14, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides documentation and DQL templates for network observability within the Dynatrace platform. It does not contain any executable scripts, binary files, or external dependencies. All referenced URLs point to official Dynatrace documentation or are internal Smartscape/Grail references.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted network flow data (IPs, ports, and protocols) from various external sources. While this represents a data ingestion surface, the skill's purpose is to generate read-only DQL queries within a monitored environment.
  • Ingestion points: Data enters through the default_network_flows bucket (OneAgent), logs via OpenTelemetry Collector (NetFlow), and logs via Amazon Data Firehose (AWS).
  • Boundary markers: The skill does not explicitly define LLM boundary markers for the data, but relies on the structured nature of DQL.
  • Capability inventory: The skill provides templates for data retrieval and aggregation but does not perform file writes, shell execution, or outbound network operations itself.
  • Sanitization: The instructions emphasize strict type casting (e.g., toIp(), toLong(), toSmartscapeId()) which serves as a validation layer when the agent interpolates values into queries.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 14, 2026, 03:44 PM
Security Audit — agent-trust-hub — dt-obs-network-flows