email-list-segmentation
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill and its included evaluation benchmarks describe services that ingest and process external subscriber data, creating a potential surface for indirect prompt injection if the data contains malicious instructions.
- Ingestion points: The evaluation tasks (e.g.,
evals/klaviyo-segment-sync-and-unsubscribe-web/task.md) rely on processing JSON data containing subscriber-controlled fields such asemailand behavioraltags. - Boundary markers: There are no explicit instructions within the skill to implement boundary delimiters or warnings to ignore embedded content when the agent processes these external data fields.
- Capability inventory: The logic described in the skill involves capabilities to interact with the Klaviyo API (network write) and generate validation reports (file write).
- Sanitization: The skill does not provide guidance for sanitizing or escaping string-based subscriber properties against potential prompt injection attacks during common workflows like sync or segment scoring.
Audit Metadata