brida-reflex-customer-support-ticket-triage
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches live availability, configuration, and agent contract information from the vendor's infrastructure, specifically checking files like /reflex/llms.txt, /reflex/agent.md, and /reflex/agent.json to verify hosted service status.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted customer support ticket data to determine handling branches.
- Ingestion points: Ticket state data including issue summaries, product areas, customer impact, and urgency descriptions identified in SKILL.md and custom-reflex.json.
- Boundary markers: Decision outputs are strictly constrained to declared branches (escalate, queue, self_serve) with explicit criteria defined in the custom-reflex.json contract.
- Capability inventory: The skill interacts with external integration surfaces via MCP, released SDKs, or REST APIs as specified in the implementation instructions.
- Sanitization: The instructions mandate the use of deterministic host code for parsing, policy enforcement, and authorization to ensure the agent remains within its authority boundary.
Audit Metadata