radarr
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: Risk of indirect prompt injection through movie metadata processing.
- Ingestion points: Movie names, collection titles, and details are fetched from the Radarr API and TMDB lookups inside
scripts/radarr.sh. - Boundary markers: Absent; untrusted text descriptions from movie metadata are rendered without encapsulation or formatting constraints.
- Capability inventory:
scripts/radarr.shcontains capabilities to add, monitor, or delete entries and files inside the local Radarr application instance via API calls. - Sanitization: Query arguments are sanitized for safe web request insertion using
jq -sRr @uri, but data returned from external lookups is interpreted without strict text validation filters.
Audit Metadata