news-summary
Pass
Audited by Gen Agent Trust Hub on Apr 6, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches news data from reputable international news organizations including BBC, Reuters, NPR, and Al Jazeera.
- [EXTERNAL_DOWNLOADS]: Accesses the official OpenAI API to generate voice briefings.
- [PROMPT_INJECTION]: The skill ingests third-party data from external RSS feeds, creating a surface for indirect prompt injection.
- Ingestion points: RSS feed content retrieved from international news sites as described in SKILL.md.
- Boundary markers: None provided in the prompts to distinguish feed content from instructions.
- Capability inventory: Use of curl for networking, sed and grep for text processing, and local file writing for audio storage.
- Sanitization: Employs a sed pattern to remove HTML tags from the retrieved RSS content.
Audit Metadata