retro-dv-home-video
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the agent to read
references/cases.mdand use tools to inspect finished media and case summaries from external domains such asgoodcase.ai,media.goodcase.ai, andx.com. - [INDIRECT_PROMPT_INJECTION]: The skill's workflow depends on ingesting and processing data from external sources to draft video prompts, which could be used to influence agent behavior.
- Ingestion points: The agent reads
references/cases.mdand fetches external content from the linked websites at runtime. - Boundary markers: The instructions mandate a structured "reference contract" (Preserve, Replace, Avoid) to guide the agent's drafting process based on the ingested evidence.
- Capability inventory: The agent generates text-based video prompt artifacts and uses browser or media inspection tools to process remote content.
- Sanitization: No explicit content sanitization, filtering, or validation is mentioned for the data fetched from external domains.
Audit Metadata