brida-reflex-sponsor-segment-detection

Pass

Audited by Gen Agent Trust Hub on Sep 21, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to retrieve live availability and configuration data from the author's hosted files, specifically /reflex/llms.txt, /reflex/agent.md, and /reflex/agent.json. These resources are used to verify SDK and API status before implementation.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted media transcripts and metadata to classify content.
  • Ingestion points: Untrusted media segments and transcripts are ingested through the mediaContext, segmentText, and nearbyDisclosure fields defined in references/custom-reflex.json and sourced from the user's environment.
  • Boundary markers: The skill relies on structured JSON fields for data ingestion rather than explicit text delimiters within the prompt instructions.
  • Capability inventory: The skill restricts the agent to semantic recommendations and explicitly prohibits protected actions such as payments, merges, account mutations, or destructive mutations.
  • Sanitization: No explicit sanitization or filtering logic is specified for the ingested media text in the provided instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 21, 2026, 08:21 PM
Security Audit — agent-trust-hub — brida-reflex-sponsor-segment-detection